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 rules/hrithik-gavankar/brainstack/team-braingit clone --depth 1 https://github.com/Hrithik-Gavankar/brainstackWhat 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 | $0.01055 | $0.01055 |
| Opus 5 | $0.00528 | $0.00528 |
| Sonnet 5 | $0.00211 | $0.00211 |
| Haiku 4.5 | $0.00105 | $0.00105 |
Grade A, and why
team-brain 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 — 111 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Team Brain — sync mode (mandatory)
If this workspace has .team-brain/ or the user is working on a shared Jira initiative / spike / epic, treat Team Brain as the crew’s shared memory.
Personal BRAIN.md / engineer-brain stay private — never upload those.
Compliance check (MCP-first, soft gate)
Policy: stronger_prompts (not a hard CLI block). Agents must follow it; humans may still use the CLI offline.
EACH TURN on a known Jira key
→ prefer MCP: sync_status / compliance / prepare_research
→ if compliance.agent_action is set → do that BEFORE deep research or ending the turn
→ research_ok must be true before digging into the codebase
(start loads context; else prepare_research / recall)
→ after durable findings → remember(source_ref) in the same turn
CLI mirror:
bash <brainstack>/core/scripts/team-brain-api.sh compliance <JIRA-KEY>
bash <brainstack>/core/scripts/team-brain-api.sh sync-status <JIRA-KEY>
Human entry (one manual step)
When the user starts team work on a ticket — including phrases like:
I'm starting on AAP-81423 — start Team Brain sync.
I'm starting on AAP-81423 — start Team Brain sync, summarize crew memory, then help me.
/team-brain start AAP-81423
Wake Team Brain sync for AAP-81423 and continue.
Stop Team Brain sync for AAP-81423.
…run the matching command (start / wake / stop):
bash <brainstack>/core/scripts/team-brain-api.sh start <JIRA-KEY>
That loads crew memory into .team-brain/cache/<KEY>.json and starts background pull.
Then you MUST:
- Read the cache /
sync-status(checkcompliance.research_ok) - Summarize crew knowledge (2–5 bullets)
- Only then dig into the codebase
While sync mode is active
EACH TURN on this key
→ touch <KEY> (keeps sync awake)
→ prefer cache; recall / prepare_research if topic-specific search needed
AFTER durable finding
→ remember with stable source_ref (KEY#slug)
→ identical body = no-op; same source_ref + new body = UPDATE (merge)
→ never invent a second row for the same topic slug
→ body style: natural prefer/avoid guidance — never TODO/NO-TODO dumps
ON HUMAN CORRECTION (bad research / wrong sync summary)
→ treat as ground truth; do not argue
→ correct(source_ref, corrected_body) or re-remember same source_ref → UPDATE
→ optional learning at source_ref/learning (was wrong → prefer …)
→ prior body archived; history/restore for soft rollback if needed
LONG SESSION (do not spam every turn)
→ after a long research block, OR about every 8–10 turns on this key
→ quiet recall / prepare_research (or summarize notify/<KEY>.json if updated)
→ optional once-per-spike: suggest background `watch <KEY> &` (poll or --push)
→ at most one soft nudge per stretch — never nag every turn
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 · 111 lines · 1,055 tokens per session scan A de412cc7d061
team-brain is a cursor rule published in the GitHub repository Hrithik-Gavankar/brainstack (5 stars, last pushed 12d ago), licensed MIT. It adds 1,055 tokens to every session, about $0.0053 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 cursor rules, from other repositories
cursorrules
Read and follow all instructions in INSTRUCTIONS.md at the repository root.
rolemule-core
Core RoleMule conventions — app name, error system, JSON serialization, background tasks.
analytics-consent-onboarding
PostHog analytics, GDPR cookie consent, and onboarding tour usage.
read-xlsx
Reading, writing, diffing, and repairing spreadsheets (.xlsx) for AI agents via the xfa MCP server.
cursorrules
Cursor rule "cursorrules" from me2resh/agent-decision-record, covering agent decision records (agdr) — cursor rules, purpose, see: https://github.com/me2resh/agent-decision-record, decision detection and adding a new dependency → stop, document decision first.
agdr
Agent Decision Records - enforces structured documentation of technical decisions before implementation.