agentleFS
Sign inStart free
vidit19sharmavidit19sharma/neogents/skills/evolve/SKILL.md

evolve

Skill1 starsChanged 11 days agoMITView on GitHub

Turn accumulated friction into one bounded, human-approved improvement to NEO itself. Reads FRICTION.md, clusters recurring patterns, proposes a single edit to an agent prompt, skill, or template — with evidence — and applies it only after explicit approval. Use when friction entries pile up or the user asks "why does NEO keep doing X".

41 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use instinct-export, instinct-import and 9 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • TheKREATORSSkill

    instinct-export

    Export instincts from project/global scope to a file

    Also in 4 projects that use this

  • TheKREATORSSkill

    instinct-import

    Import instincts from file or URL into project/global scope

    Also in 4 projects that use this

  • diego-tobalinaSkill

    build-fix

    Fix build and compilation errors systematically. Run when build fails.

    Also in 3 projects that use this

  • launch52-aiSkill

    skill-create

    Create and validate Claude Code skills with proper structure, checklists, and validation scripts. Use when creating a new skill, auditing existing skills, or ensuring skills meet quality standards. Generates complete skill scaffolds with SKILL.md, check scripts, guides, and templates.

    Also in 3 projects that use this

  • qtzx06Skill

    eval

    Evaluate trained YOLO model accuracy with mAP metrics and per-class error analysis. Identifies failure cases for targeted improvement. Use after training.

    Also in 5 projects that use this

  • MontimageSkill

    test-coverage

    Improve test coverage by finding untested branches, error paths, and edge cases. Use when measuring or strengthening a project's tests. Don't use for production debugging, code review, or feature implementation.

    Also in 3 projects that use this

  • +1 more from TheKREATORS

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 36 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    MUST NEVER edit: ${CLAUDE_PLUGIN_ROOT}/hooks/ (any script or hooks.json), ${CLAUDE_PLUGIN_ROOT}/.claude-plugin/, CI …

    line 27

“Common” is measured: kept by at least 6 in 10 of the 36 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

/neo:evolve — improve NEO from evidence

Friction ledger: !if test -f .neo/brain/FRICTION.md && test ! -L .neo/brain/FRICTION.md; then head -c 1048576 .neo/brain/FRICTION.md | grep -c '^- \[F-' || true; else echo 0; fi entries

NEO's components (agent prompts, skills, templates) are files. Friction entries are evidence of where those files fail. This command closes the loop — one bounded edit at a time, never on a hunch.

Steps

  1. Guard: no .neo/brain/FRICTION.md or fewer than 3 entries → say there isn't enough evidence to evolve yet and stop. Evolution on thin evidence is guessing.
  2. Cluster: read FRICTION.md and group entries by the component they implicate. A pattern needs 2+ entries pointing at the same component — single incidents don't justify edits. Skip entries already marked (addressed by …). Cluster from FRICTION's raw entries themselves, never from a prior evolve proposal's summary of them — re-summarizing a summary compounds loss.
  3. Propose ONE edit — the most recurrent cluster only. Present an evidence manifest before touching anything:
    • Evidence: the [F-NNN] entries cited, verbatim.
    • Root cause: what in the component invites this failure — one paragraph.
    • Targeted fix: the exact edit (quote current text → proposed text). Smallest change that addresses the pattern. Prefer adding a narrow rule over rewriting sections.
    • Predicted impact: what should stop happening, and what's at risk of regressing.
  4. Editable surface (hard boundary):
    • MAY edit: ${CLAUDE_PLUGIN_ROOT}/agents/*.md, ${CLAUDE_PLUGIN_ROOT}/skills/*/SKILL.md, ${CLAUDE_PLUGIN_ROOT}/templates/** — prompt-layer only.
    • On a marketplace install ${CLAUDE_PLUGIN_ROOT} is the plugin cache, which Claude Code replaces on update: edits there are lost unless also sent upstream (PR) or applied in a --plugin-dir checkout.
    • MUST NEVER edit: ${CLAUDE_PLUGIN_ROOT}/hooks/ (any script or hooks.json), ${CLAUDE_PLUGIN_ROOT}/.claude-plugin/, CI (${CLAUDE_PLUGIN_ROOT}/.github/). Enforcement stays outside the loop it enforces.
  5. Human gate: apply the edit only after an explicit yes. On a marketplace install, tell the user before asking that the edit lands in the plugin cache and is lost on the next plugin update unless sent upstream as a PR. "Maybe" or silence means no. If rejected, have neo-shadow append the rejection as a new FRICTION entry so the next run knows it was tried.
  6. After applying: re-read the edited file to confirm the change is coherent in context, and check the edit didn't weaken an existing rule (zero-regression: an evolve edit may add or sharpen constraints, never quietly remove one). Ask shadow to record the change in DECISIONS.md, and mark the cited friction entries with (addressed by <file>, YYYY-MM-DD).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free