Skip to main content

capabilities

Your capability catalog — read this at boot. Lists the temporal date-range skills and the external integrations (reached via the loopback broker) available to you as a spawned worker, and exactly how to call each. Read-only. Consult it whenever you're unsure what tools/integrations you have or how to invoke them.

Informations de source

Dépôt
HarnessMD/munder-difflin
Dernière activité de la source
19 juin 2026 à 16:09
Langue détectée de SKILL.md
anglais
Étoiles
8 661
Forks
1 179

Options d'installation

Le prompt qui vérifie d'abord la source est sélectionné par défaut. Vous pouvez passer à une commande directe ou télécharger une copie locale.

Vérifiez les fichiers source

Lisez SKILL.md et les fichiers associés affichés par SkillsMP avant de décider de l'installer.

Affichage de SKILL.md

SKILL.md
Instructions source · Aperçu en lecture seule
name
capabilities
description
Your capability catalog — read this at boot. Lists the temporal date-range skills and the external integrations (reached via the loopback broker) available to you as a spawned worker, and exactly how to call each. Read-only. Consult it whenever you're unsure what tools/integrations you have or how to invoke them.
allowed-tools
["Bash"]
# Worker Capability Catalog You are an autonomous worker spawned by the hive to do one objective and report back. This catalog tells you **what you can do and how to call it** — your temporal skills and your external integrations — so you don't have to guess. Everything here is read-only to invoke (the integrations themselves may act, but they are mediated and credential-free from your side). ## 1. Your environment The harness injects these env vars (use them; don't hard-code paths): - `AGENT_ID`, `AGENT_NAME` — your identity in the hive. - `AGENT_DIR` — your private workspace (`identity.md`, `memory.md`, `inbox/`, `outbox/`, and `.claude/skills/`). Your bundled skills live under `$AGENT_DIR/.claude/skills/`. - `HIVE_ROOT` — the shared hive (`PROTOCOL.md`, the kanban, other agents). At boot you also have `identity.md` (who you are) and `HIVE_ROOT/PROTOCOL.md` (the full coordination protocol). To message god or another agent, write ONE message JSON into `$AGENT_DIR/outbox/` (schema in PROTOCOL.md). When finished, send god an `"act":"done"` outbox message with a substantive result summary. ## 2. Temporal skills — concrete date ranges, relative to now When your task is time-scoped, resolve the dates instead of computing them by hand. Each skill prints inclusive civil dates (`YYYY-MM-DD`, your local timezone) **and** the half-open `[startUtc, endExclusiveUtc)` instants for timestamp queries. All are read-only (clock + stdout only; no writes, no network). Named shortcuts (invoke directly): | Skill | Range it resolves | | --- | --- | | `/today`, `/yesterday` | the single civil day | | `/thisWeek`, `/lastWeek` | ISO week (Mon-start); this = Mon→today, last = prior full week | | `/last7Days`, `/last30Days` | rolling N-day window ending today | | `/thisMonth`, `/lastMonth` | this = 1st→today; last = prior full month | | `/thisQuarter`, `/lastQuarter` | this = quarter-start→today; last = prior full quarter | | `/thisYear`, `/lastYear` | this = Jan 1→today (YTD); last = prior full year | For **any** window — including `last90Days`, `last12Months`, or arbitrary `lastNdays` / `lastNweeks` / `lastNmonths` — use `/temporal`, or call the resolver directly: ```bash node "$AGENT_DIR/.claude/skills/temporal/when.mjs" last30Days # one window node "$AGENT_DIR/.claude/skills/temporal/when.mjs" # all windows node "$AGENT_DIR/.claude/skills/temporal/when.mjs" --json 90d # JSON only node "$AGENT_DIR/.claude/skills/temporal/when.mjs" --list # keywords ``` Convention: `this*` windows are **period-start → today** (to-date); `last*` named periods are the **full prior complete period**. ## 3. Integrations — via the loopback broker External services are reached through the hive's **loopback broker**: a local, authenticated endpoint the harness runs on `127.0.0.1`. You call the broker; it holds the credentials and performs the outbound call on your behalf. You never see or store a secret, and only loopback callers are accepted — so integration access is brokered, auditable, and credential-free from your side. How integrations are surfaced to you is **environment-dependent** — the exact set enabled depends on how you were spawned and the hive's configuration — so discover what's live at run time rather than assuming. The broker pattern is stable even as specific integrations are added; treat the list below as the current surface, not a fixed contract. Concrete capabilities you may have right now (check availability before relying on one): - **Reply to the originating Slack thread.** If your objective arrived from Slack, your dispatch includes the exact loopback reply command — `node "<helper>" --channel <C> --thread <T> --text "<mrkdwn>"`. Use *that* command verbatim to post your result back to the thread. Send a substantive reply (a short `*bold*` headline + the actual outcome/links), never a bare "done". - **Hive messaging.** Coordinate or hand off by writing an outbox message JSON to god or another agent (`$AGENT_DIR/outbox/`). This is your always-available channel. - **Semantic memory (MemPalace).** If enabled for you: `mempalace search "<query>"` to recall shared knowledge, `mempalace wake-up` for a digest. - **Enterprise Knowledge Graph.** If enabled: `node "$KG_CLI" search "<query>"` for ranked passages, `node "$KG_CLI" list`, `node "$KG_CLI" get <id>` — use it for company-specific facts instead of guessing. - **MCP integrations** (filesystem, git, and others) come pre-wired into your session settings when enabled; invoke them as normal tools. The set is gated by the hive's consent configuration. As additional brokered integrations land (calendar, mail, docs, web fetch, …), they follow the same shape: a brokered, credential-free call discoverable at run time. Pair them with the temporal skills above — resolve the date window first, then pass those concrete ISO bounds to the integration query. ## 4. Boundaries - Temporal skills are **read-only** helpers — they never write or reach the network. The broker mediates all external calls; you hold no credentials. - Do **not** push or tag to any remote. Commit locally; **god** is the sole integrator. Pause only for high-severity actions (remote push, paid/infra changes, deleting something you didn't create) — otherwise work autonomously. - Finish by reporting to god (`"act":"done"`) with a real, substantive summary.
Voir sur GitHub