hivemind-goals — Claude Code skill: team goals and tasks
This skill tracks team goals and parked tasks through the hivemind CLI, so they stay visible to the whole team instead of sitting in local files.
npx skills add activeloopai/hivemind --skill hivemind-goals -a claude-codeWhat hivemind-goals does
The skill hands the agent a set of hivemind goal commands: create a goal, list them, close one, change its status. Mention a goal, an objective or a "remind me to" item and the agent runs the CLI and reports back the goal_id. From there the goal is visible to the team in Deeplake.
Parked tasks get their own path. When a side task shows up mid-session, the skill stores it with enough context to resume cold: a "Start here" line, files, branch, run command and the reason. The whole package goes in as one double-quoted argument, so newlines survive. On resume, the agent pulls the full package with hivemind goal get <goal_id> and picks up where it left off.
One hard limit: on the Hermes runtime goals must go through the shell CLI. Writing files under ~/.deeplake/memory/goal/... lands on the local filesystem and never reaches the shared hivemind_goals table, so teammates will not see those goals. The hivemind-memory skill describes a different memory layout and does not apply here.
What it helps with
- create a team goal and get its
goal_id - park a side task with context for later
- close a goal or move it to in progress
- list your own goals from the terminal
- resume a parked task from the right file
Example prompts for the agent
- add a goal: finish the billing migration
- remind me later to add rate-limiting to the webhook handler
- let's pick up the webhook task
How to install hivemind-goals
1Skills CLI (Vercel)
npx skills add activeloopai/hivemind --skill hivemind-goals -a claude-code2Manually: put the files in place
~/.claude/skills/hivemind-goals/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/hivemind-goals/
1Skills CLI (Vercel)
npx skills add activeloopai/hivemind --skill hivemind-goals -a codex2Manually: put the files in place
.agents/skills/hivemind-goals/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add activeloopai/hivemind --skill hivemind-goals -a cursor2Manually: put the files in place
.cursor/skills/hivemind-goals/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install activeloopai/hivemind hivemind-goals2Manually: put the files in place
.github/skills/hivemind-goals/SKILL.md1Skills CLI (Vercel)
npx skills add activeloopai/hivemind --skill hivemind-goals -a gemini-cli2Platform CLI
gemini skills install https://github.com/activeloopai/hivemind --path harnesses/hermes/skills/hivemind-goalsManually: put the files in place
.devin/skills/hivemind-goals/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add activeloopai/hivemind --skill hivemind-goals -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about hivemind-goals
How do I install hivemind-goals in Claude?
Run npx skills add activeloopai/hivemind --skill hivemind-goals -a claude-code. The skill file lands at ~/.claude/skills/hivemind-goals/SKILL.md.
What does hivemind-goals do?
It creates, tracks and updates team goals through the hivemind CLI, and stores parked tasks together with the context needed to resume them.
Why not just write a goal to a file?
On the Hermes runtime a write under ~/.deeplake/memory/goal/... stays on the local disk and never reaches the shared hivemind_goals table, so the team will not see it.
Source text
SKILL.md · Apache-2.0
--- name: hivemind-goals description: Create, track and update team goals in Hivemind via the `hivemind` CLI. Use whenever the user mentions a goal, objective, target, milestone, or asks to track progress on something measurable. ALSO use when the user says "task", "todo", "work item", "remind me to", "fix X", or any actionable work item — the goal system replaced the legacy `hivemind tasks` CLI and now covers both objectives and tasks. allowed-tools: terminal --- # Hivemind Goals — CLI only (Hermes) ⚠️ **CRITICAL: On this runtime (Hermes), you MUST use the `hivemind` shell CLI for goals. DO NOT use `write_file` on `~/.deeplake/memory/goal/...` paths — those writes go to the local filesystem and never reach the team-shared `hivemind_goals` table. Other team members will NOT see them.** The hivemind-memory skill describes a generic memory layout — it does NOT apply to goals. For goals, use the CLI below. ## Commands (invoke via terminal tool) ``` hivemind goal add "<text>" # create goal, prints goal_id hivemind goal list [--mine|--all] # list (default --mine) hivemind goal done <goal_id> # mark closed hivemind goal progress <goal_id> <opened|in_progress|closed> ``` ## Workflow when the user expresses a goal 1. `hivemind goal add "<short description>"` — capture stdout, that's the `goal_id` (UUID). 2. Tell the user the goal_id and that it is now team-visible in Deeplake. ## Capture a task for later (with resumable context) When the user **parks a tangential task** mid-session — "save this for later", "remind me to …", "don't let me forget …", "let's do X later" — store enough **context to resume cold** later, not just a one-liner. Tag it `--agent capture` so parked side-tasks are separable from hand-made goals: ``` hivemind goal add --agent capture "Add rate-limiting to the webhook handler Start here: add a per-IP token bucket on the handler entry path Files: src/webhook/handler.ts:120-160, src/webhook/limits.ts Branch: feat/webhook-hardening Run: pnpm test webhook Why: bursty clients hammer the endpoint; defer until retry-backoff lands" ``` Line 1 is the label (what `goal list` shows). Fill `Start here / Files / Branch / Run / Why` from the conversation; `Start here:` matters most. Pass the whole package as **one double-quoted argument** so the newlines are preserved. ## Resume a parked task (automatic context transfer) When the user says "let's work on that task / goal" or "pick up the `<X>` task": 1. `hivemind goal list --mine` — match the user's reference to a `goal_id`. 2. `hivemind goal get <goal_id>` — prints the **full** package (`goal list` shows only the first line, so always use `goal get`). Read it as your working context. 3. `hivemind goal progress <goal_id> in_progress` — mark it started. 4. Begin from `Start here:` using the `Files` / `Branch` / `Run` lines. Continue as if the context was never lost. ## What NOT to do - Do NOT call `write_file` on any path under `~/.deeplake/memory/goal/`. - Do NOT do `mkdir` / `cat >` to create those files manually via terminal. If the user wants to inspect goals you created, run `hivemind goal list --mine` (terminal) and present the output.
© activeloopai · Apache-2.0 · activeloopai/hivemind