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 agentmods add commands/lollipopkit/cc-plugins/cf-rungit clone --depth 1 https://github.com/lollipopkit/cc-pluginsWhat 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 | $0.00024 | $0.01084 |
| Opus 5 | $0.00012 | $0.00542 |
| Sonnet 5 | $0.00005 | $0.00217 |
| Haiku 4.5 | $0.00002 | $0.00108 |
Grade A, and why
cf-run 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.
How it starts
The opening of the file, as written. The whole thing — 123 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Run the context-firewall workflow.
Arguments
$ARGUMENTS
Expected:
--spec <file|json>(required)--out <path>(optional)
Steps
1) Resolve project root
- Run
pwdvia Bash and treat it as the project root directory.
2) Load TaskSpec JSON (--spec)
-
Extract the
--specvalue from$ARGUMENTS.- If parsing fails or
--specis missing, emit a schema-valid SubResult.v1 withstatus: "tool_error"and a singlefollowupsarray containing the required usage message (e.g.,--spec <file|json> [--out <path>]).
- If parsing fails or
-
If the
--specvalue, after trimming leading whitespace, begins with{, treat it as inline JSON. -
Otherwise treat it as a file path (use the untrimmed value):
- If it is not absolute, resolve to an absolute path by joining with the project root.
- Read the file content via Read.
3) Parse and validate TaskSpec.v1
- Parse the spec content as JSON.
- Validate minimum fields:
version == "TaskSpec.v1"task_idis non-emptyoutput_schema == "SubResult.v1"inputsis a non-empty arrayquestionsis a non-empty array
Error handling (MANDATORY, JSON-only):
- If JSON parsing/validation fails, output a schema-valid SubResult.v1 JSON:
version: "SubResult.v1"task_id: use"unknown"if missingstatus: "tool_error"answers: []coverage.inputs_scanned: []followups: describe the missing/invalid fields
4) Load settings (optional)
- If
<root>/.claude/context-firewall.local.mdexists, read it and parse YAML frontmatter. - If frontmatter has
enabled: false, output a schema-valid SubResult.v1 JSON with:status: "partial"answers: []coverage.inputs_scanned: []followups: [{"need":"plugin disabled","suggested_next_task":"set enabled: true in .claude/context-firewall.local.md"}]
5) Persist TaskSpec
- Determine persistence root:
- Default:
<root>/.claude/context-firewall/ - If settings frontmatter has
persist.dir, use that path (resolve relative to<root>). - If settings frontmatter has
persist.enabled: false, skip all persistence steps.
- Default:
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.
- 2d ago First seen · 123 lines · 0 tokens per session scan A 24881e603d65
cf-run is a command published in the GitHub repository lollipopkit/cc-plugins (7 stars, last pushed 5mo ago), licensed MIT. It adds 24 tokens to every session and 1,084 once invoked, about $0.0001 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.
Other commands, from other repositories
advisor
Get a second opinion from Codex (GPT-5) on your current plan, diff, or a specific question.
triage
Triage review findings one by one — present pending TODOs for user decision (approve, skip, modify).
constatar-run
Execute a constatar plan.toml and report grounded results.
constatar-audit
Audit a constatar run — conformance, health trends, cost.
constatar-verify
Verify a unit of work with grounded evidence via constatar.
remoto-status
Show the remote agent fleet: host health and all jobs with their states.