How do I install the Extract Build Log skill?

Run npx skillmds@latest add https-deeplearning-ai/extract-build-log in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Extract Build Log safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Extract Build Log?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Extract Build Log free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Extract Build Log?

https-deeplearning-ai (@https-deeplearning-ai) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Extract Build Log by https-deeplearning-ai

Name: Extract Build Log
Author: https-deeplearning-ai

Extract Build Log

Extracts the build conversation for a recorded run in the spec-build-lab and writes it to evals/<run>/session-log.md. Use this skill ANY time the user wants to capture, save, snapshot, extract, export, or render the conversation for a specific build run — including phrasings like "extract the build log", "save the conversation for run-02", "snapshot run-02", "export the build log", "render the session log", "log this build", "produce the session log for run-NN", or "/extract-build-log run-NN". Reads the transcript recorded by Claude Code (located via the .session breadcrumb at evals/<run>/.session) and slices it between an opening bookend (the /prepare-build announcement) and a closing bookend (a fuzzy diagram/structure phrase by default, or a user-supplied --until pattern). Idempotent — re-runnable with different bookends. Requires an explicit run-NN; if none is given, list the runs in builds/ and ASK which one — never default. Do NOT use this skill to evaluate the run — that's /eval-spec-vs-build or /eval-m

File contents

extract-build-log

Slice the build conversation out of Claude Code's session transcript and render it as a markdown log at evals/<run>/session-log.md.

Steps

Verify cwd is a course folder. Confirm spec.md, materials/, builds/, evals/ exist. Refuse otherwise: "Run /extract-build-log from inside a course folder — cd courses/<name> first." Stop.
Require a run. If $ARGUMENTS (or its first whitespace-separated token) is empty:
```
ls -1 builds/ 2>/dev/null | grep -E '^run-[0-9]+$' | sort
```
List those runs and ask: "Which run should I extract? Pick one — I won't default to the latest." Stop until they answer.
Validate the run. Confirm evals/$ARGUMENTS/.session exists. Refuse otherwise: "evals/$ARGUMENTS/.session is missing — this run wasn't allocated by /prepare-build, so I can't locate its transcript." Stop.
Run the extractor. From the project root:
```
python3 "$CLAUDE_PROJECT_DIR/.claude/skills/extract-build-log/scripts/extract.py" \
  "evals/$ARGUMENTS" \
  [--until="<user-supplied phrase>"]
```
Pass --until only if the user gave one. The script:
- Reads evals/$ARGUMENTS/.session to find the transcript
- Slices between the opening bookend (the /prepare-build announcement matching this run number) and the closing bookend (--until regex if given, else a fuzzy default, else end of transcript)
- Writes evals/$ARGUMENTS/session-log.md (overwrites if present — re-run freely with different bookends)
Report. Echo the script's stdout — it prints the path written plus a one-line summary of how many turns were extracted and which bookends were used.
Hand off. Tell the user: "Ready to evaluate run-$ARGUMENTS — I can run /eval-spec-vs-build $ARGUMENTS and /eval-materials-vs-build $ARGUMENTS next. Say 'go ahead' to proceed, or pick one if you only want to run one of them."

Don't

Don't generate or judge content — this skill is a transcript slicer, not an analyst.
Don't read or write to builds/<run>/ — that's the build artifact, out of scope.
Don't pick a default run. List the runs in builds/ and ask.
Don't fabricate a .session file if it's missing — that means the run wasn't allocated normally; refuse and ask the user to investigate.

---name: extract-build-logdescription: Extracts the build conversation for a recorded run in the spec-build-lab and writes it to evals/<run>/session-log.md. Use this skill ANY time the user wants to capture, save, snapshot, extract, export, or render the conversation for a specific build run — including phrasings like "extract the build log", "save the conversation for run-02", "snapshot run-02", "export the build log", "render the session log", "log this build", "produce the session log for run-NN", or "/extract-build-log run-NN". Reads the transcript recorded by Claude Code (located via the .session breadcrumb at evals/<run>/.session) and slices it between an opening bookend (the /prepare-build announcement) and a closing bookend (a fuzzy diagram/structure phrase by default, or a user-supplied --until pattern). Idempotent — re-runnable with different bookends. Requires an explicit run-NN; if none is given, list the runs in builds/ and ASK which one — never default. Do NOT use this skill to evaluate the run — that's /eval-spec-vs-build or /eval-m---# extract-build-logSlice the build conversation out of Claude Code's session transcript andrender it as a markdown log at `evals/<run>/session-log.md`.## Steps1. **Verify cwd is a course folder.** Confirm `spec.md`, `materials/`, `builds/`, `evals/` exist. Refuse otherwise: "Run /extract-build-log from inside a course folder — `cd courses/<name>` first." Stop.2. **Require a run.** If `$ARGUMENTS` (or its first whitespace-separated token) is empty: ```bash ls -1 builds/ 2>/dev/null | grep -E '^run-[0-9]+$' | sort ``` List those runs and ask: "Which run should I extract? Pick one — I won't default to the latest." Stop until they answer.3. **Validate the run.** Confirm `evals/$ARGUMENTS/.session` exists. Refuse otherwise: "evals/$ARGUMENTS/.session is missing — this run wasn't allocated by /prepare-build, so I can't locate its transcript." Stop.4. **Run the extractor.** From the project root: ```bash python3 "$CLAUDE_PROJECT_DIR/.claude/skills/extract-build-log/scripts/extract.py" \ "evals/$ARGUMENTS" \ [--until="<user-supplied phrase>"] ``` Pass `--until` only if the user gave one. The script: - Reads `evals/$ARGUMENTS/.session` to find the transcript - Slices between the opening bookend (the `/prepare-build` announcement matching this run number) and the closing bookend (`--until` regex if given, else a fuzzy default, else end of transcript) - Writes `evals/$ARGUMENTS/session-log.md` (overwrites if present — re-run freely with different bookends)5. **Report.** Echo the script's stdout — it prints the path written plus a one-line summary of how many turns were extracted and which bookends were used.6. **Hand off.** Tell the user: "Ready to evaluate run-$ARGUMENTS — I can run /eval-spec-vs-build $ARGUMENTS and /eval-materials-vs-build $ARGUMENTS next. Say 'go ahead' to proceed, or pick one if you only want to run one of them."## Don't- Don't generate or judge content — this skill is a transcript slicer, not an analyst.- Don't read or write to `builds/<run>/` — that's the build artifact, out of scope.- Don't pick a default run. List the runs in `builds/` and ask.- Don't fabricate a `.session` file if it's missing — that means the run wasn't allocated normally; refuse and ask the user to investigate.

Frequently asked questions