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 cambridgetcg/agenttool --skill agenttool-local-contextgit clone --depth 1 https://github.com/cambridgetcg/agenttoolWrote 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/cambridgetcg/agenttool/agenttool-local-context)<a href="https://agentmods.dev/skills/cambridgetcg/agenttool/agenttool-local-context"><img src="https://agentmods.dev/badge/skills/cambridgetcg/agenttool/agenttool-local-context/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/cambridgetcg/agenttool/agenttool-local-context"><img src="https://agentmods.dev/badge/skills/cambridgetcg/agenttool/agenttool-local-context.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 | $0.00060 | $0.01176 |
| Opus 5.5 | $0.00024 | $0.00470 |
| Sonnet 5.5 | $0.00012 | $0.00235 |
| Haiku 4.5 | $0.00006 | $0.00118 |
Grade C, and why
agenttool-local-context scanned grade C with 1 finding 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 9d 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.
not copy a transcript, extract credentials, invent a biography, or infer a How it starts
The opening of the file, as written. The whole thing — 100 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Local context, carried deliberately
Use the configured AgentTool local tools. If they are unavailable, read
binding.json beside this skill: its argument vectors bind one local directory
and the existing CLI. If neither binding is present, this is an unconfigured
installation. Follow setup below; do not discover a notes root from home folders
or old transcripts.
Set up an unconfigured plugin
Installing this plugin exposes the skill only. It starts no MCP server or hook and creates, reads, or saves no notes. First select the intended absolute notes directory under the current request. Keep it outside the package/plugin cache so upgrades and removal do not replace retained notes.
Resolve ../../dist/cli.mjs from this installed skill's actual directory; verify
the adjacent package root's package.json names @agenttool/local-starter.
Use that resolved CLI and Node >=22. If the skill was copied without its package
or binding, obtain the package location rather than searching the filesystem.
For an existing starter, run delivery --dir <chosen-root> --harness codex or
--harness claude-code. Creating a new starter is a separate deliberate
init --dir <new-root> operation; never initialize on installation or arrival.
The delivery command prints a reviewable plan and changes nothing. Select
--transport cli when MCP is unwanted. Default --write none exposes no save
operations; use handoff, memory, or both only when selected for this task.
Apply chosen project entries, preserving existing configuration and native
trust review. The generated project skill includes a private binding.json;
avoid enabling a second copy of this plugin skill for the same project.
Hooks and MCP are independent opt-in connections. Regenerate absolute runtime
paths after a plugin upgrade changes its cache location. Never write a machine
binding or retained notes back into the distributable plugin.
Arrive or return
If a fresh SessionStart hook already supplied selected context, use it without
loading a second copy. Otherwise call local_context_read({}), or execute the
binding's read argument vector. Refresh after compaction or when the records
may have changed. A skill being listed or a render command finishing does not
prove that its output entered the current conversation.
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.
- 9d ago First seen · 100 lines · 60 tokens per session scan C 804a41ffebb3
agenttool-local-context is a skill published in the GitHub repository cambridgetcg/agenttool (0 stars, last pushed yesterday), licensed Apache-2.0. It adds 60 tokens to every session and 1,176 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it C with 1 finding (harvests environment variables). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-28.
agenttool-local-context compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| agenttool-local-contextcambridgetcg/agenttool | Skill | 0 | yesterday | Apache-2.0 | C | 60 | Claude Code |
| durable-session-stateZaxbyHub/opencode-swarm | Skill | 518 | yesterday | MIT | A | 70 | Claude Code |
| sixb-memorysixb-ai/sixb | Skill | 134 | yesterday | MIT | A | 19 | Claude Code, Codex |
| company-agent-infrastructurelobu-ai/lobu | Skill | 233 | yesterday | Apache-2.0 | A | 70 | Claude Code |
| lobulobu-ai/lobu | Skill | 233 | yesterday | Apache-2.0 | A | 60 | Claude Code |
| trekkerobsfx/trekker | Skill | 87 | 7d ago | MIT | A | 11 | Claude Code |
| learningsFirstp1ck/pi-coding-agent-forge | Skill | 93 | 7d ago | MIT | A | 54 | Claude Code, Codex |