rgit-capture

A workflow for capturing and organizing recent repository changes into research records. It uses rgit, a tool that keeps searchable proposals about code changes.

In plain words
What is it for?
Use it after an rgit run, commit, or watch process leaves proposals open, or when recent changes need to be segmented and saved.
Why use it?
It prevents open change proposals from becoming unorganized or impossible to recall and query later.

Skill for Claude CodeCodex

Install

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.

agentmods
npx agentmods add skills/stepzerolab/research-git/rgit-capture
Any agent
npx skills add StepzeroLab/research-git --skill rgit-capture
Clone the repo
git clone --depth 1 https://github.com/StepzeroLab/research-git

Made for: Claude Code, Codex.

Per session 71 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,569 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What 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.

ModelPer sessionOnce invoked
Fable 5 $0.00071 $0.01569
Opus 5 $0.00036 $0.00785
Sonnet 5 $0.00014 $0.00314
Haiku 4.5 $0.00007 $0.00157

Measured 2d ago against content hash e0001e0b8c91, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

rgit-capture 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.

src/rgit/_plugin/skills/rgit-capture/SKILL.md · 97 lines

How it starts

The opening of the file, as written. The whole thing — 97 lines — stays where its author put it; the contents beside it link to each section on GitHub.

rgit-capture

Orchestrates the two-phase capture: a free, deterministic Phase 1 through the rgit CLI, then an agentic Phase 2 dispatched natively onto the host session's subscription — no paid API.

Prerequisites: the target repo has been rgit init-ed.

Locating the agent definitions. On Claude Code the plugin runtime resolves agent paths for you. On other CLIs (Codex, Gemini, opencode) this skill is symlinked into ~/.agents/skills/rgit-capture, so resolve the plugin root once and reference the agents from there:

SKILL_REAL=$(realpath ~/.agents/skills/rgit-capture 2>/dev/null || readlink -f ~/.agents/skills/rgit-capture)
PLUGIN_ROOT=$(dirname "$(dirname "$SKILL_REAL")")    # the bundled _plugin/ directory

Every agents/<name>.md reference below (agents/capsule-segmenter.md, agents/edge-judge.md) lives at $PLUGIN_ROOT/agents/<name>.md.

Process

1. Ensure there are proposals to segment (Phase 1 — free, deterministic)

If the user just made changes and there is no open proposal yet:

rgit capture                 # picks for you: uncommitted work, or the last commit when the tree is clean
rgit capture main..HEAD      # a specific span of commits (any A..B range)

Repeated captures of the same diff dedup into the existing proposal, and repos with the post-commit hook (rgit install-hooks) capture each commit automatically — don't capture the same commit twice. Proposals also appear from rgit run and the rgit watch daemon.

2. Read the pending captures

Run rgit pending --json → a list of {proposal_id, trigger, diff, candidates}. The diff is the raw material; the candidates are crude heuristic guesses you are about to replace. If the list is empty, tell the user there is nothing to segment and stop.

3. Dispatch the capsule-segmenter subagent (Phase 2 — agentic, on subscription)

For each pending proposal, dispatch a subagent using the capsule-segmenter agent definition (agents/capsule-segmenter.md); run independent proposals concurrently. Pass in the dispatch prompt: proposal_id, repo_root (absolute path of the target repo), diff (verbatim from rgit pending), and symbols if available. The subagent returns {"capsules": [...], "dropped": [...]} — high-quality capsules with real intent / knobs / data_assumptions / resurrection_guide, infrastructure noise dropped.

Read the full file on GitHub · 97 lines

Changes

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.

  1. 2d ago First seen · 97 lines · 71 tokens per session scan A e0001e0b8c91

Subscribe to this mod's changes

rgit-capture is a skill published in the GitHub repository StepzeroLab/research-git (42 stars, last pushed 26d ago), licensed MIT. It adds 71 tokens to every session and 1,569 once invoked, about $0.0004 per session on Opus 5. 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.

Related

Other skills, from other repositories

ts-sdk-author

Design, build, verify, and publish production-grade TypeScript SDKs as npm packages inside a pnpm monorepo. Covers workspace layout, public API and module boundaries, plugin extension points, branded types and library-tuned tsconfig, tsdown bundling (vs tsup/tsc-only/unbuild), package.json exports with dual ESM+CJS…

mindfold-ai/Trellis · 219 tokens

first-principles-thinking

Systematic first principles thinking for any problem domain. Use when the user says "analyze from first principles", "第一性原理", "从根本分析", "从零开始思考", "think from scratch", "question this design", "is this the right approach", "challenge assumptions", "挑战假设", "为什么要这样做", "有没有更好的方案", "why are we doing it this way", or needs…

mindfold-ai/Trellis · 164 tokens

python-design

Python design patterns for CLI scripts and utilities — type-first development, deep modules, complexity management, and red flags. Use when reading, writing, reviewing, or refactoring Python files, especially in .trellis/scripts/ or any CLI/scripting context. Also activate when planning module structure, deciding…

mindfold-ai/Trellis · 73 tokens

contribute

Guide for contributing to Trellis documentation and marketplace. Covers adding spec templates, marketplace skills, documentation pages, and submitting PRs across both the Trellis main repo and docs repo. Use when someone wants to add a new spec template, add a new skill to the marketplace, add or update documentation…

mindfold-ai/Trellis · 72 tokens

trellis-brainstorm

Guides collaborative requirements discovery before implementation. Creates task directory, seeds PRD, asks high-value questions one at a time, researches technical choices, and converges on MVP scope. Use when requirements are unclear, there are multiple valid approaches, or the user describes a new feature or complex…

mindfold-ai/Trellis · 65 tokens

trellis-meta

Understand and customize the local Trellis architecture inside a user project. Use when modifying .trellis plus platform hooks, settings, agents, skills, commands, prompts, workflows, the channel runtime (trellis channel), bundled runtime agents under .trellis/agents/, selectable workflow templates, registry-backed…

mindfold-ai/Trellis · 110 tokens