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.
npx skills add rapidreview-io/Merv --skill git-accessgit clone --depth 1 https://github.com/rapidreview-io/MervWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/rapidreview-io/merv/git-access)<a href="https://agentmods.dev/skills/rapidreview-io/merv/git-access"><img src="https://agentmods.dev/badge/skills/rapidreview-io/merv/git-access/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/rapidreview-io/merv/git-access"><img src="https://agentmods.dev/badge/skills/rapidreview-io/merv/git-access.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00049 | $0.03804 |
| Opus 5.5 | $0.00020 | $0.01522 |
| Sonnet 5.5 | $0.00010 | $0.00761 |
| Haiku 4.5 | $0.00005 | $0.00380 |
Grade C, and why
git-access scanned grade C with 3 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 16d 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.
Harvests environment variableshighData exfiltration
Enumerating or grepping the environment for keys collects credentials unrelated to what the mod says it does.
env = dict((k, v) for k, v in os.environ.items() Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
from urllib.parse import urlsplit Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
subprocess.run(["openssl", "genpkey", "-algorithm", "RSA", "-pkeyopt", How it starts
The opening of the file, as written. The whole thing — 269 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Git through the signed-in Merv account
For code-linked project work, perform this synchronization throughout the active task. Use the existing native MCP connection; do not send the person to Settings/Keys or ask them for a token. List projects and choose the intended explicit projectId, asking only if ambiguous. Read its brief and code status. Git is optional for non-code research tasks. A reader/reviewer can clone or fetch; producer/operator users can push ordinary user branches. Workers use their assigned credentials, never this flow.
The code.git_credential tool takes {projectId, publicKey, mode}. Use read for
clone/fetch or write for the user-branch synchronization authorized by normal code-linked work. It returns a relative Git URL,
username merv, mode, expiresAt, encryption RSA-OAEP-SHA256, and encryptedToken.
The token expires in ten minutes and loses access immediately if the source grant
or membership is revoked. A role change invalidates existing derived credentials;
request another with the current role. Push admission and reserved refs still apply.
Never print a decrypted token, put it in a URL, command argument, environment
variable, plaintext file, tool output or chat. Do not read the native client's
private OAuth store. Only public keys and ciphertext cross the model/tool boundary.
Use Python 3, OpenSSL and Git's built-in credential-cache, as below. These are
ordinary local commands, not an installed Merv helper. Do not substitute Git's
plaintext credential store or enable tracing. If these commands are unavailable,
report the missing prerequisite; do not create a long-lived key as a workaround.
Keep code-linked work synchronized while you work
For a code-linked project, synchronization is the default part of normal work, not another setup step. Use the selected project's repository and current intended user branch. If the repository/project mapping or branch is ambiguous, resolve that ambiguity before uploading; do not upload an unrelated working directory. In a fresh chat, rediscover the branch and remote instead of inheriting another chat's selection. Each agent keeps its own intended branch; never switch to another agent's branch merely because it is newer. Readers/reviewers fetch and report local-only changes; producer/operator members can push user branches through the same login.
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.
- 16d ago First seen · 269 lines · 49 tokens per session scan C a409d62d36e4
git-access is a skill published in the GitHub repository rapidreview-io/Merv (7 stars, last pushed 4d ago), licensed Apache-2.0. It adds 49 tokens to every session and 3,804 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it C with 3 findings (harvests environment variables, makes network calls, runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-22.
git-access compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| git-accessrapidreview-io/Merv | Skill | 7 | 4d ago | Apache-2.0 | C | 49 | Claude Code |
| merge-batchdiegosouzapw/OmniRoute | Skill | 85k | today | MIT | A | 46 | Claude Code |
| prowler-commitprowler-cloud/prowler | Skill | 15k | yesterday | Apache-2.0 | A | 44 | Claude Code, Codex |
| gh-auth-isolationgithub/gh-aw | Skill | 5.6k | today | MIT | A | 20 | Claude Code, Codex |
| comet-githubrpamis/comet | Skill | 3.3k | today | MIT | A | 72 | Claude Code, Codex |
| github-skillzeenie-ai/OpenCompany | Skill | 1.1k | 3d ago | MIT | A | 51 | Claude Code |
| re0-mergeLilMGenius/paperthin | Skill | 1.4k | today | MIT | A | 70 | Claude Code |