Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add bdfinst/agentic-dev-team/plugin install dev-teamWrote 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/bdfinst/agentic-dev-team/headless-run)<a href="https://agentmods.dev/skills/bdfinst/agentic-dev-team/headless-run"><img src="https://agentmods.dev/badge/skills/bdfinst/agentic-dev-team/headless-run/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/bdfinst/agentic-dev-team/headless-run"><img src="https://agentmods.dev/badge/skills/bdfinst/agentic-dev-team/headless-run.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 | $0.00132 | $0.01378 |
| Opus 5.5 | $0.00053 | $0.00551 |
| Sonnet 5.5 | $0.00026 | $0.00276 |
| Haiku 4.5 | $0.00013 | $0.00138 |
Grade C, and why
headless-run 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 1mo 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.
(`make_cell_home` / `cell_env` / `dispatch`), which does `env = dict(os.environ)` How it starts
The opening of the file, as written. The whole thing — 108 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Headless Run
Role: worker. This skill runs one Claude Code prompt in a maximally isolated subprocess and returns its JSON result — it does not review, plan, or edit.
When to use it
Use it for scripted, one-shot headless invocations — above all a benchmark
harness that shells out to claude -p per case (the #821 harness that runs
/code-review once per fixture). A plain nested claude -p launched inside a
live Claude Code Remote session inherits the parent's identity and tool surface
(see docs/using-plugin-skills-in-the-web-environment.md and issue #842):
- Reused
session_id— the nested run reports the same session id as the parent, so runs are not independent. - Remote-injected tool surface — the nested run sees Remote-runtime MCP tools a local CLI session never exposes, changing the tool set under test.
This skill is the reusable workaround for that #842 constraint.
Isolation guarantees
The helper builds each dispatch to leak as little parent state as possible:
- Fresh session id. A new
uuid.uuid4()is passed as--session-id <uuid>, so the nested run cannot reuse the parentsession_id. This is the key fix. - Clean HOME + config. A throwaway temp
HOMEandCLAUDE_CONFIG_DIR— no shared config,.claude/memory/, or telemetry state carries over. - Scrubbed env. Inherited
CLAUDE_*session/Remote identity vars (e.g.CLAUDE_SESSION_ID,CLAUDE_CODE_SESSION_ID,CLAUDE_CODE_ENTRYPOINT,CLAUDE_CODE_REMOTE) are removed, thenIS_SANDBOX=1andDEV_TEAM_TELEMETRY=offare set fresh. - No resume. The command never uses
--resume/-ror--fork-session. - JSON result + timeout. Runs with
--output-format jsonand a hard subprocess timeout; prints the verifiedsession_id, cost, and token usage.
Auth vs. isolation (#957)
The fresh HOME also wipes Claude Code's login state, so a dispatch
reports "Not logged in" unless ANTHROPIC_API_KEY is set. ~/.claude.json
alone does not fix this — confirmed empirically, twice — something
under ~/.claude/ itself gates the login check ahead of Claude Code's
actual (Keychain-backed) OAuth token lookup. Pass --preserve-auth to
copy ~/.claude.json and most of ~/.claude/ into the cell home first,
restoring login for operators who authenticate via claude login rather
than an API key. This is a real tradeoff, not free: it carries over
settings.json, projects/, sessions/, mcpServers, plugins/, and
other app state — copy_auth_state()'s _CLAUDE_DIR_EXCLUDE skips the
clearly bulky, clearly-not-auth-related pieces (history.jsonl,
file-history/, session-env/, paste-cache/, shell-snapshots/,
debug/, telemetry/, downloads/), but everything else rides along
since we don't have a confirmed narrower answer for exactly which piece
satisfies the login check. Off by default; the code-review-benchmark
harness (runner.make_isolated_dispatch_fn()) turns it on unconditionally,
since running that harness at all presupposes the operator's own
subscription.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 1mo ago First seen · 108 lines · 132 tokens per session scan C 080af9400a03
headless-run is a skill published in the GitHub repository bdfinst/agentic-dev-team (295 stars, last pushed 4d ago), licensed MIT. It adds 132 tokens to every session and 1,378 once invoked, about $0.0005 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 2026-09-03.
Other skills, from other repositories
research-engineer
An uncompromising Academic Research Engineer. Operates with absolute scientific rigor, objective criticism, and zero flair. Focuses on theoretical correctness, formal verification, and optimal implementation across any required technology.
eval-dataset-design
A guide to creating reliable test sets for AI agents. Each evaluation task includes the request, the simulated user's behaviour, a reset starting state, and a success check that can be independently verified.
tika-eval-compare
Compare extracts from two Tika builds over a corpus to detect regressions in content, encoding, exceptions, and embedded-document handling. Use for "compare before/after extracts", "eval this change against the corpus".
cua-driver
Use Cua Driver for desktop or browser tasks that are awkward or unavailable through Bash/APIs, or when the user explicitly wants GUI interaction: app testing, visual bug reproduction, form filling, calendar entry, screenshots, and demo recording. Also covers Cua setup; not OpenAI Codex Computer Use or web research.
jetson-validate-image
Use after jetson-flash-image to run static BSP checks, on-target smoke/regression tests on a flashed DUT, or both. Not for build or flash steps. Triggers: validate bsp, on-target validation.
atmos-validation
Validate Atmos projects, components, arbitrary JSON Schema inputs, EditorConfig, and GitHub Actions; use affected-file selection and native CI annotations.