evolve-skill-from-traces SKILL.md skillA

evolve-skill-from-traces from pjt222/agent-almanac is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add pjt222/agent-almanac/evolve-skill-from-traces@v1.0

Pins v1.0 in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
pjt222/agent-almanac
Path
i18n/caveman-lite/skills/evolve-skill-from-traces/SKILL.md
Versions recorded
5
Latest version
v1.0, committed Aug 17, 2026
First version recorded
Apr 19, 2026
Size
15,955 bytes
Repository stars
43
License
MIT

Evolve SKILL.md files from agent execution traces using a three-stage pipeline: trajectory collection from observed runs, parallel multi-agent patch proposal for error and success analysis, and conflict-free consolidation of overlapping edits via prevalence-weighting. Based on the Trace2Skill methodology.

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# Evolve a Skill from Execution Traces

Transform raw agent execution traces into a validated SKILL.md through a three-stage pipeline: trajectory collection, parallel multi-agent patch proposal, and conflict-free consolidation. This skill bridges the gap between observed agent behavior and documented procedures, turning successful runs into reproducible skills.

## When to Use

- Execution traces reveal recurring patterns not captured in existing skills
- Observed agent behavior outperforms the documented procedure
- Building skills from scratch by recording expert demonstrations
- Multiple agents propose conflicting improvements to the same skill

## Inputs

- **Required**: `traces` -- set of agent execution logs or session transcripts (minimum 10 successful runs recommended)
- **Required**: `target_skill` -- path to an existing SKILL.md to evolve, or `"new"` for skill extraction from scratch
- **Optional**: `analyst_count` -- number of parallel analyst agents to spawn (default: 4)
- **Optional**: `held_out_ratio` -- fraction of traces reserved for validation, not used in drafting (default: 0.2)

## Procedure

### Step 1: Collect Execution Traces
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add pjt222/agent-almanac/evolve-skill-from-traces@sha256:0a9e6143d4efc105

The line at the top of this page pins the label v1.0, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_luro6comxxuwef7q.svg)](https://markdownregistry.com/a/art_luro6comxxuwef7q)

3 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 5 versions of this file, between Apr 19, 2026 and Aug 17, 2026, about one every 30 days.

versioncommittedcommitsizeaudit
v1.0 latest2026-08-17 a784899 15,955 BA view · diff
v1.02026-06-05 33b561c 15,926 BA view · diff
v1.02026-06-05 75ded7a 15,902 BA view · diff
v1.02026-04-24 0b91b04 15,944 BA view · diff
v1.02026-04-19 e334ebd 16,021 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (15955 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

pjt222/agent-almanac · 43 stars · license MIT · pushed 2026-11-06 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_luro6comxxuwef7q
GET https://markdownregistry.com/api/v1/resolve?ref=pjt222/agent-almanac/evolve-skill-from-traces
GET https://markdownregistry.com/api/v1/blob/0a9e6143d4efc1050a22984fc5d6b26c39cc2b32de4b0cb3f9de9182aaa3c423

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from pjt222/agent-almanac

CLAUDE.md CLAUDE.md file
pjt222/agent-almanac · CLAUDE.md
git:20261105.d2f5ad6 · audit A · 43 stars
adapt-architecture SKILL.md skill
pjt222/agent-almanac · i18n/caveman-lite/skills/adapt-architecture/SKILL.md · Execute structural metamorphosis using strangler fig migration, chrysalis phases, and interface preservation. Covers…
v1.0 · audit A · 43 stars
adaptic SKILL.md skill
pjt222/agent-almanac · i18n/caveman-lite/skills/adaptic/SKILL.md · Master skill composing the 5-step synoptic cycle for panoramic synthesis across multiple domains. Orchestrates…
v1.0 · audit A · 43 stars
add-puzzle-type SKILL.md skill
pjt222/agent-almanac · i18n/caveman-lite/skills/add-puzzle-type/SKILL.md · Scaffold a new puzzle type across all 10+ pipeline integration points in jigsawR. Creates the core puzzle module, wires…
v1.0 · audit A · 43 stars
add-rcpp-integration SKILL.md skill
pjt222/agent-almanac · i18n/caveman-lite/skills/add-rcpp-integration/SKILL.md · Add Rcpp or RcppArmadillo integration to an R package for high-performance C++ code. Covers setup, writing C++…
v1.0 · audit A · 43 stars
aikido SKILL.md skill
pjt222/agent-almanac · i18n/caveman-lite/skills/aikido/SKILL.md · Practice aikido as a defensive martial art emphasizing harmony, redirection, and controlled resolution. Covers…
v1.0 · audit A · 43 stars
analyze-codebase-for-mcp SKILL.md skill
pjt222/agent-almanac · i18n/caveman-lite/skills/analyze-codebase-for-mcp/SKILL.md · Analyze an arbitrary codebase to identify functions, APIs, and data sources suitable for exposure as MCP tools…
v1.0 · audit A · 43 stars
analyze-codebase-workflow SKILL.md skill
pjt222/agent-almanac · i18n/caveman-lite/skills/analyze-codebase-workflow/SKILL.md · Analyze an arbitrary codebase to auto-detect workflows, data pipelines, and file dependencies using putior's put_auto()…
v1.0 · audit A · 43 stars
analyze-diffusion-dynamics SKILL.md skill
pjt222/agent-almanac · i18n/caveman-lite/skills/analyze-diffusion-dynamics/SKILL.md · Analyze the dynamics of diffusion processes using stochastic differential equations, Fokker-Planck equations…
v1.0 · audit A · 43 stars
analyze-generative-diffusion-model SKILL.md skill
pjt222/agent-almanac · i18n/caveman-lite/skills/analyze-generative-diffusion-model/SKILL.md · Analyze pre-trained generative diffusion models (Stable Diffusion, DALL-E, Flux) by computing quality metrics (FID, IS…
v1.0 · audit A · 43 stars
analyze-kernel-bottleneck SKILL.md skill
pjt222/agent-almanac · i18n/caveman-lite/skills/analyze-kernel-bottleneck/SKILL.md · Systematically identify whether a GPU kernel is compute-bound, memory-bound, or latency-bound using roofline analysis…
v1.0 · audit A · 43 stars
analyze-magnetic-field SKILL.md skill
pjt222/agent-almanac · i18n/caveman-lite/skills/analyze-magnetic-field/SKILL.md · Calculate and visualize magnetic fields produced by current distributions using the Biot-Savart law, Ampere's law, and…
v1.0 · audit A · 43 stars

Every file in pjt222/agent-almanac

Browse by kind, by grade A, or by owner.