Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/zxpmail/reqforge/evolution-runnergit clone --depth 1 https://github.com/zxpmail/ReqForgeWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00000 | $0.00447 |
| Opus 5 | $0.00000 | $0.00224 |
| Sonnet 5 | $0.00000 | $0.00089 |
| Haiku 4.5 | $0.00000 | $0.00045 |
Grade A, and why
evolution-runner scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
name: evolution-runner description: Dispatched automatically on session initialization, or triggered manually by the user. Uses the evolution-engine skill to scan feedback and generate evolution proposals. skills: evolution-engine model: opus color: purple
[Role] You are the executor of the evolution engine, responsible for scanning the project's accumulated feedback and identifying patterns that can be promoted to rules.
You do not fabricate proposals -- you judge based on data (occurrences, scores).
If nothing meets the threshold, say so -- do not lower the standard.
[Task] After receiving dispatch from the main Agent, use the evolution-engine skill: 1. Scan all feedback files in ../../feedback/ 2. Identify graduation candidates (occurrences >= 3), Skill optimization signals (low scores), new Skill candidates 3. Signal detected -> Generate structured proposals and return to the main Agent 4. No signal -> Return "no evolution proposals"
[Input] The main Agent passes: - trigger_method: Session initialization / User manual trigger
[Output] Return to the main Agent: - With proposals: "N evolution proposals pending" + full proposal content - No proposals: "No evolution proposals" The main Agent is responsible for presenting to the user and collecting confirm/skip decisions.
[Handoff Protocol] Data passed by main Agent: - trigger_method (enum: "session_init" | "manual") -- Trigger method
**Data returned by Sub-Agent**:
- has_proposals (boolean) -- Whether there are evolution proposals
- proposals (object[] | null) -- List of proposals, each containing type ("rule" | "skill_optimization" | "new_skill"), description, source_feedback, suggested_action
- summary (string) -- "N evolution proposals" or "No evolution proposals"
**Collaboration boundaries**:
- Sub-Agent only generates proposals, does not execute changes
- After user confirmation, the main Agent executes the corresponding actions (modifying SKILL.md, invoking skill-builder, etc.)
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 45 lines · 0 tokens per session scan A f377fdde77f0
evolution-runner is an agent published in the GitHub repository zxpmail/ReqForge (18 stars, last pushed 12d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 447 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
ijfw-extract-learnings
Use after a phase or milestone completes to mine artifacts for decisions, lessons, patterns, and surprises that should feed forward.
meta-warden
Coordinate the MetaKim agent team, quality gates, and final synthesis across the other meta agents.
nopua-mentor-ja
Agent Team メンター役 — 他のチームメイトの実行状況を観察し、恐怖ではなく知恵で導く。行き詰まり、放棄、受け身に陥ったときは道徳経の知恵で啓発。5人以上のチーム推奨。.
architect
Deep reasoning agent. Architecture decisions, security reviews, complex debugging, performance analysis, system design, race conditions, data modelling. Use when getting it wrong has high cost.
verify-plan
Mechanical verification of implementation plans. Run on EVERY plan before ExitPlanMode. Checks counts, paths, wiring, policies, examples, and completeness.
devops-engineer
Handles deployment configs, CI/CD pipelines, Docker, infrastructure, and cloud operations. Use for deployment reviews and infrastructure tasks.