skills.rest
contexthub-md
contexthub-mdOFFICIAL
Explore all skills by this author →
GitHub

mission-feedback

Debriefs completed missions by analyzing Claude Code session transcripts into scored findings.

Updated May 12, 2026
Search Tags:#claude code#retrospective#post-mortem#scoring#transcript analysis#session review#mission debrief
One-click install
npx skills add https://github.com/contexthub-md/marketplace --skill mission-feedback-contexthub-md
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mission-feedback
Source: https://github.com/contexthub-md/marketplace/tree/main/contexthub/skills/mission-feedback
Command: npx skills add https://github.com/contexthub-md/marketplace --skill mission-feedback-contexthub-md

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? After a development mission finishes, it is hard to know what actually went wrong or what to improve next time. This Skill reads the real Claude Code session transcript for a mission and turns it into an evidence-based debrief with scores and actionable findings, instead of relying on subjective recollection. ## Core Features & Use Cases - Transcript Resolution: Maps a mission ID (full UUID, weaver-<short>, or bare short ID) to its git worktree and the encoded ~/.claude/projects/ transcript folders, with fallback scanning. - Run Reconstruction: Parses .jsonl transcript files (excluding agent- sidechains) to rebuild the goal, tool calls, errors, retries, and per-model token usage. - Scored Debrief: Rates flow, quality, agenticness, and PM on a 1-10 scale with evidence-backed rationale, then produces a ranked findings list classified as bugfix, prompt-change, or skill-creation. - Use Case: After a mission like weaver-02d2080b completes, run the debrief to get a dated markdown report in contexthub-docs/feedback/ with concrete follow-up actions, routed only after human approval. ## Quick Start Ask the assistant to run mission-feedback on mission weaver-02d2080b to generate a transcript-driven debrief with scores and ranked findings.

Frequently Asked Questions about mission-feedback

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I debrief a completed Claude Code mission?▼

Run the mission-feedback skill with the mission ID, such as /contexthub:mission-feedback weaver-02d2080b. It resolves the mission to its transcript folder, reads the session JSONL files, and produces a scored debrief with ranked findings.

How does mission-feedback differ from the reflect skill?▼

The reflect skill is subjective and manual, capturing end-of-run product and process notes. Mission-feedback is transcript-driven: it reads the actual session record, scores four dimensions, and cites specific turns as evidence for every finding.

What mission ID formats are accepted?▼

It accepts a full UUID, a weaver-<short> form, or a bare 8-character short ID. The short ID is derived by taking the first 8 characters of the UUID or stripping the weaver- prefix, then matched against worktree transcript folders.

What happens if no transcript folder matches the mission ID?▼

The skill falls back to scanning all of ~/.claude/projects/ for any folder whose name contains the short ID, since the mission may have run in a non-worktree workspace. It confirms with the user before proceeding.

Does mission-feedback modify transcripts or write to Brain automatically?▼

No. It is strictly read-only against ~/.claude/projects/ and never modifies transcripts. All routing of findings, such as filing bug reports or editing CLAUDE.md, requires explicit human approval first.

How are findings classified in the debrief?▼

Each finding is classified as exactly one of bugfix (a defect in tooling), prompt-change (missing guidance the agent needed), or skill-creation (a repeatable workflow worth codifying), with a concrete proposed action and turn-level evidence.

Related Skills

Explore More Software Engineering Skills→
its-Basudebaits-Basudeba
1

claude-devfleet

Orchestrate parallel AI agents in isolated git worktrees via MCP.

Community
Advanced
GavinTPAGavinTPA

lets-wrap

Capture session learnings and commit approved edits to mission-context and CLAUDE.md.

Community
Intermediate
xurikxurik
1

dw-skill-eval-score

Score DataWorks Skill transcripts via rule checks, LLM judgments, and statistics.

Community
Advanced
Soul-Brews-StudioSoul-Brews-Studio
116

retrospective

Capture retrospective notes and generate concise lessons-learned summaries via /rrr alias.

Official
Basic