Skip to content
Back to search

context-engineering-marketplace

A comprehensive collection of Agent Skills for context engineering, multi-agent architectures, and production agent systems. Use when building, optimizing, or debugging agent systems that require effective context management.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/muratcankoylan-agent-skills-for-context-engineering/badge.svg)](/listings/muratcankoylan-agent-skills-for-context-engineering)

HTML

<a href="/listings/muratcankoylan-agent-skills-for-context-engineering"><img src="/api/listings/muratcankoylan-agent-skills-for-context-engineering/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

**Step 1: Add the Marketplace**

Run this command in Claude Code to register this repository as a plugin source:

/plugin marketplace add muratcankoylan/Agent-Skills-for-Context-Engineering

**Step 2: Install the Plugin**

Option A - Browse and install: 1. Select `Browse and install plugins` 2. Select `context-engineering-marketplace` 3. Select `context-engineering` 4. Select `Install now`

Option B - Direct install via command:

/plugin install context-engineering@context-engineering-marketplace

This installs all published skills in a single plugin. Skills are activated automatically based on your task context.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 39 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. {
  2. "name": "context-engineering-marketplace",
  3. "owner": {
  4. "name": "Muratcan Koylan",
  5. "email": "muratcan.koylan@outlook.com"
  6. },
  7. "metadata": {
  8. "description": "Context engineering and harness engineering skills plus a file-based autonomous research-to-skill operating system with measured router-benchmark results across four frontier models",
  9. "version": "2.6.0"
  10. },
  11. "plugins": [
  12. {
  13. "name": "context-engineering",
  14. "description": "Comprehensive context engineering and harness engineering skills for production-grade AI agent systems: fundamentals, degradation patterns, compression, optimization, self-managed context (model-controlled live context editing), multi-agent coordination, long-horizon prompting (pseudo-formal task briefs for long-running and parallel agents), memory systems, tool design, filesystem context, hosted agents, evaluation, autonomous harnesses, self-improvement loops (RSI, meta-harnesses, evolutionary scaffold search), latent briefing (KV cache sharing between agents), project development, and cognitive architecture. Ships with a researcher operating system (rubrics, mechanism registry, claim provenance, run state machine, adversarial benchmarks, continuous loop).",
  15. "source": "./",
  16. "strict": false,
  17. "skills": [
  18. "./skills/context-fundamentals",
  19. "./skills/context-degradation",
  20. "./skills/context-compression",
  21. "./skills/context-optimization",
  22. "./skills/self-managed-context",
  23. "./skills/multi-agent-patterns",
  24. "./skills/long-horizon-prompting",
  25. "./skills/memory-systems",
  26. "./skills/tool-design",
  27. "./skills/filesystem-context",
  28. "./skills/hosted-agents",
  29. "./skills/evaluation",
  30. "./skills/advanced-evaluation",
  31. "./skills/harness-engineering",
  32. "./skills/project-development",
  33. "./skills/bdi-mental-states",
  34. "./skills/latent-briefing",
  35. "./skills/self-improvement-loops"
  36. ]
  37. }
  38. ]
  39. }
README.md · README · 364 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. # Agent Skills for Context Engineering
  2. A comprehensive, open collection of Agent Skills focused on context engineering and harness engineering principles for building production-grade AI agent systems. These skills teach the art and science of curating context, designing agent operating loops, and evaluating agent behavior across any agent platform.
  3. [![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/muratcankoylan/Agent-Skills-for-Context-Engineering)
  4. ## What is Context Engineering?
  5. Context engineering is the discipline of managing the language model's context window. Unlike prompt engineering, which focuses on crafting effective instructions, context engineering addresses the holistic curation of all information that enters the model's limited attention budget: system prompts, tool definitions, retrieved documents, message history, and tool outputs.
  6. The fundamental challenge is that context windows are constrained not by raw token capacity but by attention mechanics. As context length increases, models exhibit predictable degradation patterns: the "lost-in-the-middle" phenomenon, U-shaped attention curves, and attention scarcity. Effective context engineering means finding the smallest possible set of high-signal tokens that maximize the likelihood of desired outcomes.
  7. ## Recognition
  8. This repository is cited in academic research as foundational work on static skill architecture:
  9. > "While static skills are well-recognized [Anthropic, 2025b; Muratcan Koylan, 2025], MCE is among the first to dynamically evolve them, bridging manual skill engineering and autonomous self-improvement."
  10. 1. [Meta Context Engineering via Agentic Skill Evolution](https://arxiv.org/pdf/2601.21557), Peking University State Key Laboratory of General Artificial Intelligence (2025)
  11. 2. [Agent Harness Engineering: A Survey](https://openreview.net/pdf/f358711a95aaaf61fdeffd4ef3fc60fba9b8da57.pdf), CMU, Yale, JHU, NEU, Tulane, UAB, OSU, Virginia Tech, and Amazon (2026)
  12. ## Skills Overview
  13. ### Foundational Skills
  14. These skills establish the foundational understanding required for all subsequent context engineering work.
  15. | Skill | Description |
  16. |-------|-------------|
  17. | [context-fundamentals](skills/context-fundamentals/) | Understand what context is, why it matters, and the anatomy of context in agent systems |
  18. | [context-degradation](skills/context-degradation/) | Recognize patterns of context failure: lost-in-middle, poisoning, distraction, and clash |
  19. | [context-compression](skills/context-compression/) | Design and evaluate compression strategies for long-running sessions |
  20. ### Architectural Skills
  21. These skills cover the patterns and structures for building effective agent systems.
  22. | Skill | Description |
  23. |-------|-------------|
  24. | [multi-agent-patterns](skills/multi-agent-patterns/) | Master orchestrator, peer-to-peer, and hierarchical multi-agent architectures |
  25. | [long-horizon-prompting](skills/long-horizon-prompting/) | **NEW** Write pseudo-formal task briefs for long-running autonomous agents and parallel orchestrations: exact success predicates, non-counting outcomes, audit-gated return conditions, per-worker handoff contracts, effort floors, and diversity policies, modeled on the published GPT-5.6 Sol Ultra Cycle Double Cover prompt and Sakana AI's MASS workflow contracts |
  26. | [memory-systems](skills/memory-systems/) | Design short-term, long-term, and graph-based memory architectures |
  27. | [tool-design](skills/tool-design/) | Build tools that agents can use effectively |
  28. | [filesystem-context](skills/filesystem-context/) | Use filesystems for dynamic context discovery, tool output offloading, and plan persistence |
  29. | [hosted-agents](skills/hosted-agents/) | **NEW** Build background coding agents with sandboxed VMs, pre-built images, multiplayer support, and multi-client interfaces |
  30. ### Operational Skills
  31. These skills address the ongoing operation and optimization of agent systems.
  32. | Skill | Description |
  33. |-------|-------------|
  34. | [context-optimization](skills/context-optimization/) | Apply compaction, masking, and caching strategies |
  35. | [self-managed-context](skills/self-managed-context/) | **NEW** Give the model read-write control over its own live context: context as an editable file, harness invariants (pinned prefix, edit gate, receipts, budget readouts, rollback), edit-position cache cost and suffix cache reuse, and steering or training context-editing strategy, based on Context Language Models |
  36. | [latent-briefing](skills/latent-briefing/) | Share task-relevant orchestrator state with workers via task-guided KV cache compaction when the worker runtime is controllable |
  37. | [evaluation](skills/evaluation/) | Build evaluation frameworks for agent systems |
  38. | [advanced-evaluation](skills/advanced-evaluation/) | Master LLM-as-a-Judge techniques: direct scoring, pairwise comparison, rubric generation, and bias mitigation |
  39. | [harness-engineering](skills/harness-engineering/) | Design autonomous agent harnesses with locked metrics, durable logs, novelty gates, rollback, and human approval boundaries |
  40. | [self-improvement-loops](skills/self-improvement-loops/) | **NEW** Build loops where the harness itself is the optimization target: RSI, meta-harness search, failure-driven self-edits, evolutionary scaffold search, and acceptance gates for self-modifying systems |
  41. ### Development Methodology
  42. These skills cover the meta-level practices for building LLM-powered projects.
  43. | Skill | Description |
  44. |-------|-------------|
  45. | [project-development](skills/project-development/) | Design and build LLM projects from ideation through deployment, including task-model fit analysis, pipeline architecture, and structured output design |
  46. ### Cognitive Architecture Skills
  47. These skills cover formal cognitive modeling for rational agent systems.
  48. | Skill | Description |
  49. |-------|-------------|
  50. | [bdi-mental-states](skills/bdi-mental-states/) | **NEW** Transform external RDF context into agent mental states (beliefs, desires, intentions) using formal BDI ontology patterns for deliberative reasoning and explainability |
  51. ## Design Philosophy
  52. ### Progressive Disclosure
  53. Each skill is structured for efficient context use. At startup, agents load only skill names and descriptions. Full content loads only when a skill is activated for relevant tasks.
  54. ### Platform Agnosticism
  55. These skills focus on transferable principles rather than vendor-specific implementations. The patterns work across Claude Code, Cursor, and any agent platform that supports skills or allows custom instructions.
  56. ### Conceptual Foundation with Practical Examples
  57. Scripts and examples demonstrate concepts using Python pseudocode that works across environments without requiring specific dependency installations.
  58. ## Usage
  59. ### Usage with Claude Code
  60. This repository is a **Claude Code Plugin Marketplace** containing context engineering skills that Claude automatically discovers and activates based on your task context.
  61. ### Installation
  62. **Step 1: Add the Marketplace**
  63. Run this command in Claude Code to register this repository as a plugin source:
  64. ```
  65. /plugin marketplace add muratcankoylan/Agent-Skills-for-Context-Engineering
  66. ```
  67. **Step 2: Install the Plugin**
  68. Option A - Browse and install:
  69. 1. Select `Browse and install plugins`
  70. 2. Select `context-engineering-marketplace`
  71. 3. Select `context-engineering`
  72. 4. Select `Install now`
  73. Option B - Direct install via command:
  74. ```
  75. /plugin install context-engineering@context-engineering-marketplace
  76. ```
  77. This installs all published skills in a single plugin. Skills are activated automatically based on your task context.
  78. ### Skill Activation Scenarios
  79. | Skill | Activate When |
  80. |-------|---------------|
  81. | `context-fundamentals` | Establishing context-window mental models, planning agent architecture, or explaining how context components affect model behavior |
  82. | `context-degradation` | Diagnosing attention failures, context poisoning, lost-in-middle behavior, or degraded agent performance across long sessions |
  83. | `context-compression` | Preserving useful state while reducing conversation, tool-output, or trajectory size under context pressure |
  84. | `context-optimization` | Improving token efficiency, retrieval precision, prefix reuse, masking, partitioning, or budget allocation for agent systems |
  85. | `self-managed-context` | Letting the model decide what stays in its live context: context-as-a-file editing, self-editing harness invariants, mid-context edit cost under prefix caching, budget readouts and nudges, and steering or training context-editing strategy |
  86. | `latent-briefing` | Sharing orchestrator trajectory with workers via task-guided KV cache compaction when the worker runtime is controllable and the models are compatible |
  87. | `multi-agent-patterns` | Choosing coordination patterns, isolating context across agents, designing handoffs, or evaluating whether parallel agents are justified |
  88. | `long-horizon-prompting` | Writing or evaluating the launch prompt for a long-running autonomous agent or parallel orchestration: success predicates, non-counting outcomes, persistence and stop rules, adversarial audit gates, per-worker output contracts, and portfolio diversity policies |
  89. | `memory-systems` | Persisting cross-session knowledge, tracking entities over time, choosing memory frameworks, or designing retrieval and update semantics |
  90. | `tool-design` | Defining agent-tool contracts, consolidating tool surfaces, improving descriptions, or making tool errors actionable |
  91. | `filesystem-context` | Moving large or durable context into files, creating scratchpads, supporting just-in-time discovery, or coordinating agents through shared artifacts |
  92. | `hosted-agents` | Running coding agents in remote sandboxes, background environments, warm pools, or multiplayer agent infrastructure |
  93. | `evaluation` | Creating deterministic checks, rubrics, regression suites, production monitoring, or quality gates for agent behavior |
  94. | `advanced-evaluation` | Using LLM judges, pairwise comparison, calibration, bias mitigation, or human-aligned quality assessment |
  95. | `harness-engineering` | Designing autonomous loops with locked evaluators, editable surfaces, durable logs, novelty gates, rollback, and approval boundaries |
  96. | `self-improvement-loops` | Building loops that modify themselves: failure-driven harness self-edits, meta-harness search, evolutionary scaffold search, context mechanism evolution, and acceptance gates for self-modification |
  97. | `project-development` | Deciding whether an LLM is appropriate, shaping batch pipelines, creating staged artifacts, or estimating operational cost |
  98. | `bdi-mental-states` | Modeling beliefs, desires, intentions, rational action traces, or neuro-symbolic state transformations for agents |
  99. <img width="1014" height="894" alt="Screenshot 2025-12-26 at 12 34 47 PM" src="https://github.com/user-attachments/assets/f79aaf03-fd2d-4c71-a630-7027adeb9bfe" />
  100. ### For Cursor, Codex, and Open Plugins
  101. This repository ships as an [Open Plugins](https://open-plugins.com) plugin. Hosts discover skills from the repo-root `skills/` directory (each subdirectory contains a `SKILL.md` file). The manifest lives at `.plugin/plugin.json`.
  102. **Cursor (recommended):**
  103. 1. Install from the [Cursor Plugin Directory](https://cursor.directory/plugins/context-engineering), or clone this repo and point Cursor at the plugin root.
  104. 2. Cursor reads `.plugin/plugin.json` and discovers the repo-root `skills/` directory through the Open Plugins manifest.
  105. 3. For project-local manual installs, copy skill directories into `.cursor/skills/`. Do not rely on repository symlinks; they are fragile on Windows and in plugin packaging.
  106. **Codex / GitHub Copilot CLI / other Open Plugins hosts:**
  107. 1. Clone or add this repository as a plugin directory.
  108. 2. The host reads `.plugin/plugin.json` and discovers all published skills under `skills/`.
  109. 3. For project-local manual installs, copy skill directories into `.codex/skills/` or the host's documented Agent Skills directory.
  110. ### Using Individual Skills
  111. Agent Skills require a **directory layout**, not a flat markdown file. Copy the skill folder into your project's skills directory:
  112. ```bash
  113. # Example: add just the context-fundamentals skill to a Cursor project
  114. mkdir -p .cursor/skills
  115. cp -R skills/context-fundamentals .cursor/skills/
  116. # Claude Code project-scoped install (same directory layout)
  117. mkdir -p .claude/skills
  118. cp -R skills/context-fundamentals .claude/skills/
  119. # Codex project-scoped install
  120. mkdir -p .codex/skills
  121. cp -R skills/context-fundamentals .codex/skills/
  122. # Generic Agent Skills repo-scoped install (Codex/OpenAI, Copilot CLI, Open Plugins hosts)
  123. mkdir -p .agents/skills
  124. cp -R skills/context-fundamentals .agents/skills/
  125. ```
  126. Do not flatten `SKILL.md` into a single file at `.claude/skills/context-fundamentals.md`. That breaks relative `references/` paths and violates the Agent Skills directory spec used by Cursor, Claude Code, and Codex.
  127. Available skills: `context-fundamentals`, `context-degradation`, `context-compression`, `context-optimization`, `self-managed-context`, `latent-briefing`, `multi-agent-patterns`, `long-horizon-prompting`, `memory-systems`, `tool-design`, `filesystem-context`, `hosted-agents`, `evaluation`, `advanced-evaluation`, `harness-engineering`, `self-improvement-loops`, `project-development`, `bdi-mental-states`
  128. ### For Custom Implementations
  129. Extract the principles and patterns from any skill and implement them in your agent framework. The skills are deliberately platform-agnostic.
  130. ## Examples
  131. The [examples](examples/) folder contains complete system designs that demonstrate how multiple skills work together in practice.
  132. | Example | Description | Skills Applied |
  133. |---------|-------------|----------------|
  134. | [digital-brain-skill](examples/digital-brain-skill/) | **NEW** Personal operating system for founders and creators. Complete Claude Code skill with 6 modules, 4 automation scripts | context-fundamentals, context-optimization, memory-systems, tool-design, multi-agent-patterns, evaluation, project-development |
  135. | [x-to-book-system](examples/x-to-book-system/) | Multi-agent system that monitors X accounts and generates daily synthesized books | multi-agent-patterns, memory-systems, context-optimization, tool-design, evaluation |
  136. | [llm-as-judge-skills](examples/llm-as-judge-skills/) | Production-ready LLM evaluation tools with TypeScript implementation, 19 passing tests | advanced-evaluation, tool-design, context-fundamentals, evaluation |
  137. | [book-sft-pipeline](examples/book-sft-pipeline/) | Train models to write in any author's style. Includes Gertrude Stein case study with 70% human score on Pangram, $2 total cost | project-development, context-compression, multi-agent-patterns, evaluation |
  138. | [interleaved-thinking](examples/interleaved-thinking/) | Reasoning trace optimizer that captures, analyzes, and converts agent failure patterns into generated skills | evaluation, advanced-evaluation, context-degradation, harness-engineering |
  139. | [long-horizon-prompt-lab](examples/long-horizon-prompt-lab/) | Production-ready educational website: method guide, copyable task-brief template, four complete prompt rewrites, structural audits, and a caveated research/vendor reference catalog | long-horizon-prompting, harness-engineering, multi-agent-patterns, advanced-evaluation |
  140. Each example includes:
  141. - Complete PRD with architecture decisions
  142. - Skills mapping showing which concepts informed each decision
  143. - Implementation guidance
  144. ### Digital Brain Skill Example
  145. The [digital-brain-skill](examples/digital-brain-skill/) example is a complete personal operating system demonstrating comprehensive skills application:
  146. - **Progressive Disclosure**: 3-level loading (SKILL.md → MODULE.md → data files)
  147. - **Module Isolation**: 6 independent modules (identity, content, knowledge, network, operations, agents)
  148. - **Append-Only Memory**: JSONL files with schema-first lines for agent-friendly parsing
  149. - **Automation Scripts**: 4 consolidated tools (weekly_review, content_ideas, stale_contacts, idea_to_draft)
  150. Includes detailed traceability in [HOW-SKILLS-BUILT-THIS.md](examples/digital-brain-skill/HOW-SKILLS-BUILT-THIS.md) mapping every architectural decision to specific skill principles.
  151. ### LLM-as-Judge Skills Example
  152. The [llm-as-judge-skills](examples/llm-as-judge-skills/) example is a complete TypeScript implementation demonstrating:
  153. - **Direct Scoring**: Evaluate responses against weighted criteria with rubric support
  154. - **Pairwise Comparison**: Compare responses with position bias mitigation
  155. - **Rubric Generation**: Create domain-specific evaluation standards
  156. - **EvaluatorAgent**: High-level agent combining all evaluation capabilities
  157. ### Book SFT Pipeline Example
  158. The [book-sft-pipeline](examples/book-sft-pipeline/) example demonstrates training small models (8B) to write in any author's style:
  159. - **Intelligent Segmentation**: Two-tier chunking with overlap for maximum training examples
  160. - **Prompt Diversity**: 15+ templates to prevent memorization and force style learning
  161. - **Tinker Integration**: Complete LoRA training workflow with $2 total cost
  162. - **Validation Methodology**: Modern scenario testing proves style transfer vs content memorization
  163. Integrates with context engineering skills: project-development, context-compression, multi-agent-patterns, evaluation.
  164. ## Researcher Operating System
  165. The [researcher](researcher/) directory is a file-based operating system for turning external research into skill changes. It exists so this repository can act as a compounding source of truth instead of an anthology.
  166. ### Measured router-benchmark results
  167. The skill router (which decides whether the right skill gets loaded for a given task) has been benchmarked end-to-end against four frontier models via the [Cursor SDK](https://cursor.com/docs/sdk/typescript). Three full sweeps (50 prompts x 4 models x 3 replications = 600 calls each):
  168. - Baseline: [`researcher/benchmarks/router/results-published/2026-05-15.md`](researcher/benchmarks/router/results-published/2026-05-15.md)
  169. - After targeted description rewrites: [`researcher/benchmarks/router/results-published/2026-05-15-v2.md`](researcher/benchmarks/router/results-published/2026-05-15-v2.md) (includes delta-vs-baseline)
  170. - After corpus-wide hardening: [`researcher/benchmarks/router/results-published/2026-05-19.md`](researcher/benchmarks/router/results-published/2026-05-19.md) (600/600 usable records, 0 format failures)
  171. Per-skill effect size for the three skills the data flagged:
  172. | Skill | Baseline top-1 | After rewrite | Delta |
  173. | --- | --- | --- | --- |
  174. | `context-fundamentals` | 0.255 | 0.489 | +23.4pp |
  175. | `project-development` | 0.750 | 1.000 | +25pp (now perfect) |
  176. | `tool-design` | 0.729 | 0.807 | +7.8pp |
  177. Per-model top-1 accuracy after the corpus-wide hardening pass:
  178. | Model | Top-1 | Top-3 |
  179. | --- | --- | --- |
  180. | gemini-3.1-pro | 0.920 | 0.933 |
  181. | composer-2 | 0.913 | 0.947 |
  182. | gpt-5.5 | 0.913 | 0.973 |
  183. | claude-opus-4-7 | 0.840 | 0.933 |
  184. Reproduce any of these numbers exactly via the runner under `researcher/benchmarks/sdk-runner/`.
  185. ### What it includes
  186. Current counts and compatibility status are generated in the [live corpus inventory](researcher/generated/corpus-summary.md). Dated benchmark and release reports retain their original snapshot counts.
  187. - **Source registry** (`researcher/source-registry.md`): priority sources, exclusion rules, monitoring queries.
  188. - **Rubrics** (`researcher/rubrics/`): content curation, skill change, harness change, pairwise skill revision.
  189. - **Mechanism registry** (`researcher/mechanisms/registry.jsonl` + `ledgers/`): accepted behavior changes used as the primary novelty signal, with append-only accepted/rejected ledgers for institutional memory.
  190. - **Claim provenance** (`researcher/claims/index.jsonl`): provenance-tracked claims with source URL, evidence strength, volatility, and last reviewed date.
  191. - **Corpus index** (`researcher/corpus/index.json`): canonical machine-readable map of skills, activation scenarios, mechanism IDs, and claim IDs.
  192. - **Run state machine** (`researcher/runs/<run-id>/run-state.json`): `initialized -> retrieved -> evaluated -> proposed -> novelty_checked -> validated -> pr_ready -> closed`.
  193. - **Activation regression tests** (`researcher/fixtures/activation-cases.jsonl`): deterministic prompts that catch skill-boundary confusion.
  194. - **Adversarial benchmark harness** (`researcher/benchmarks/`): scenarios that try to game the loop (duplicate mechanisms, unretrieved evidence, wrong rubric math, self-approved rubric changes, weak-evidence novelty).
  195. - **Continuous loop** (`researcher/scripts/loop_*.py` + `researcher/orchestration/launchd/`): inbox, source discovery, one-state-at-a-time advancement, daily ops, parked review queue, launchd service definitions.
  196. - **Skill health gate** (`researcher/scripts/skill_health.py`): deterministic body-quality scoring. Published scores remain dated evidence; local runs produce ignored runtime reports.
  197. - **Public export boundary** (`governance/export-policy.yaml` + `researcher/scripts/validate_export.py`): allowlisted projections from private or restricted records into reviewable public staging trees without publishing private source locators or digests.
  198. - **Schema and artifact contract** (`researcher/schemas/` + `researcher/scripts/artifact_store.py`): digest-pinned cross-runtime schemas, typed IDs, private bindings, exact-byte CAS, candidate freeze receipts, and Python/TypeScript conformance.
  199. ### Operator commands
  200. Install the validation dependencies once before running local gates:
  201. ```bash
  202. python3 -m pip install -r requirements-dev.txt
  203. ```
  204. ```bash
  205. # Deterministic gates (also run in CI on every PR)
  206. python3 -m unittest researcher.scripts.tests.test_skill_frontmatter
  207. python3 researcher/scripts/validate_platform_compat.py --require-reference-validator
  208. python3 researcher/scripts/validate_repo.py --strict
  209. python3 researcher/scripts/skill_health.py --strict --no-history
  210. python3 researcher/scripts/run_benchmarks.py
  211. python3 researcher/scripts/check_activation_cases.py
  212. # Per-run readiness (active runs only)
  213. python3 researcher/scripts/validate_run.py --run-dir researcher/runs/<run-id>
  214. # Continuous loop, manual
  215. python3 researcher/scripts/loop_discover.py
  216. python3 researcher/scripts/loop_step.py --allow-fetch
  217. python3 researcher/scripts/loop_daily.py
  218. python3 researcher/scripts/loop_status.py
  219. # Continuous loop, daemon (macOS)
  220. researcher/orchestration/launchd/install.sh # install launchd jobs (10-min step, 12h discover, daily ops)
  221. researcher/orchestration/launchd/uninstall.sh # remove launchd jobs
  222. ```
  223. See [researcher/runbooks/continuous-operation.md](researcher/runbooks/continuous-operation.md) for daemon details, budgets, and the human review surface.
  224. ### Guarantees
  225. - The loop never invokes paid LLMs or makes outbound writes; HTTP retrieval is stdlib-only with a 1.5 MB cap and a 30-second timeout.
  226. - Mechanism promotion requires a recorded human reviewer and a passing run-readiness check.
  227. - All queue mutations are atomic (temp file + `os.replace`) and serialized via `fcntl` locks.
  228. - Agents may prepare PRs after gates pass; merge and push remain human-controlled.
  229. ## Star History
  230. <img width="3664" height="2808" alt="star-history-2026526" src="https://github.com/user-attachments/assets/c9f88769-21b8-4762-9472-d4cf1fe1c802" />
  231. ## Structure
  232. Each skill follows the Agent Skills specification:
  233. ```
  234. skill-name/
  235. ├── SKILL.md # Required: instructions + metadata
  236. ├── scripts/ # Optional: executable code demonstrating concepts
  237. └── references/ # Optional: additional documentation and resources
  238. ```
  239. See the [template](template/) folder for the canonical skill structure.
  240. ## Contributing
  241. This repository follows the Agent Skills open development model. Contributions are welcome from the broader ecosystem. When contributing:
  242. 1. Follow the skill template structure
  243. 2. Provide clear, actionable instructions
  244. 3. Include working examples where appropriate
  245. 4. Document trade-offs and potential issues
  246. 5. Keep SKILL.md under 500 lines for optimal performance
  247. Feel free to contact [Muratcan Koylan](https://x.com/muratcan) for collaboration opportunities or any inquiries.
  248. ## License
  249. MIT License - see LICENSE file for details.
  250. ## References
  251. The principles in these skills are derived from research and production experience at leading AI labs and framework developers. Each skill includes references to the underlying research and case studies that inform its recommendations.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.