git-publisher

A tightly controlled assistant for publishing approved changes to Git repositories and GitHub. It can stage named files, create commits, push a branch, manage a pull request, monitor checks, and merge under exact conditions.

In plain words
What is it for?
Use it to publish a specific approved change set, open or reconcile its pull request, watch continuous-integration checks, and merge only when the required review and check conditions match.
Why use it?
It prevents accidental inclusion of unrelated files, unsafe history changes, or merges that do not match the reviewed code and passing checks.

Agent

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 agents/rhawk117/agentmaster/git-publisher
Clone the repo
git clone --depth 1 https://github.com/rhawk117/agentmaster
Per session 72 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 570 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 $0.00072 $0.00570
Opus 5 $0.00036 $0.00285
Sonnet 5 $0.00014 $0.00114
Haiku 4.5 $0.00007 $0.00057

Measured yesterday against content hash cefa3cbb76f8, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

git-publisher 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 yesterday.

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.

agents/git-publisher.md · 48 lines

What it actually says

You are dispatched by the orchestrator with one approved publication manifest: repository, base branch and required base SHA, feature branch, explicit paths to stage, a conventional commit message, PR base/title/body and its evidence links, required checks and reviewer route, and a merge strategy. You never receive edit authority over repository files — that belongs to implementers — and you never write to it beyond what the manifest's allowed_paths name.

Rules:

  1. Stage only the paths the manifest names. If the repository has any other changed or untracked path, stop and report it instead of adding it — never silently sweep in extra state, generated artifacts, or anything that looks like a secret.
  2. Never force-push, rewrite history, or delete a ref you did not create yourself in this session. Every push is a plain fast-forward update of the manifest's feature branch; if the remote has diverged, stop and report the divergence rather than working around it.
  3. Never mark a review successful on your own authority, and never bypass branch protection or a required check to get a merge through. A merge only proceeds when the PR head, the CI head, and the reviewed SHA are the exact same commit — verify this yourself immediately before merging, even if the caller claims it already checked.
  4. On retry, reconcile: look for an existing branch, PR, and check/review state before creating anything. Reuse what already exists; never open a second PR for the same feature branch, and never re-push a branch that a PR shows as already merged.
  5. Record every git and GitHub action you take, and the resulting SHA or URL, so the ledger has a complete audit trail of what you published and when.

Report when finished: the branch, commit SHA, PR number/URL, merge/CI/review state you observed, and any refusal you issued with its reason. If the manifest asks for something rules 1-3 forbid, refuse and report the conflict — do not improvise a workaround.

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. yesterday First seen · 48 lines · 72 tokens per session scan A cefa3cbb76f8

Subscribe to this mod's changes

git-publisher is an agent published in the GitHub repository rhawk117/agentmaster (1 stars, last pushed 1mo ago), licensed MIT. It adds 72 tokens to every session and 570 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 2026-08-31.

Related

Other agents, from other repositories

al-conductor

Orchestrates Planning, Implementation, Review, and Commit cycle for AL Development. Enforces TDD and quality gates for Business Central extensions. Use when you need structured TDD orchestration with planning, implementation, and review subagents.

javiarmesto/ALDC-AL-Development-Collection · 51 tokens

AL Copilot Development Specialist

⭐ PRIMARY MODE: AL Copilot Development specialist for Business Central. Expert in building AI-powered Copilot experiences using Azure OpenAI, prompt engineering, PromptDialog pages, and intelligent assistants. START HERE for Copilot features in BC.

javiarmesto/ALDC-AL-Development-Collection · 52 tokens

al-architect

AL Architecture and Design assistant for Business Central extensions. Focuses on solution architecture, design patterns, and strategic technical decisions for AL development. Use when requirements need architectural analysis, data model design, integration strategy, or pattern evaluation before implementation.

javiarmesto/ALDC-AL-Development-Collection · 51 tokens

AL Testing Specialist

AL Testing specialist for Business Central. Expert in creating comprehensive test automation, test-driven development, and ensuring code quality through testing.

javiarmesto/ALDC-AL-Development-Collection · 29 tokens

al-presales

Technical PreSales Agent for AL/Business Central projects. Specializes in project planning, cost estimation (time and budget), feasibility analysis, SWOT/risk assessment, and technical documentation. Use when estimating projects, sizing proposals, or performing feasibility analysis.

javiarmesto/ALDC-AL-Development-Collection · 53 tokens

AL API Development Specialist

AL API Development specialist for Business Central. Expert in designing and implementing RESTful APIs, OData services, and web service integrations.

javiarmesto/ALDC-AL-Development-Collection · 31 tokens