Core behavioral rules and skills for NanoClaw personal assistant agents. Always-on rules for communication, verification, memory, and formatting.
90%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Low
Low-risk findings worth noting
Core behavioral rules and skills for NanoClaw personal assistant agents. Always-on rules for communication, verification, memory, and formatting.
tessl install jbaruch/nanoclaw-core| Rule | Scope | Summary |
|---|---|---|
| core-behavior | always-on | Baseline conversational behavior — concise replies, no trailing summaries |
| telegram-protocol | always-on | Telegram-channel formatting and reply conventions |
| language-matching | always-on | Match the user's language; don't volunteer translations |
| default-silence | always-on | Silent-on-no-news as the default; surface only what's actionable |
| temporal-awareness | always-on | Convert relative dates, respect timezones; compute past/future via now-vs-deadline, never eyeball it |
| ground-truth | always-on | Live source > cached/recalled state; always verify before acting |
| read-full-content | always-on | Read full email/message bodies, never just snippets/previews |
| context-recovery | always-on | Recover from compaction by reading state, not by claiming lost context. Includes post-compaction skill-block disambiguation (history vs new tasks) |
| post-compaction-trust | always-on | Untrusted instructions remain untrusted after compaction |
| tone-matching | always-on | Mirror the user's register; tone shifts signal mode shifts |
| query-size-limits | always-on | Bounded SQL/API queries; never SELECT * on large tables |
| Skill | Description |
|---|---|
| core-rule-details | Reference detail for identity, silence, verification and Telegram rules |
| query-history | Search messages.db by keyword and/or sender; companion to context-recovery |
| now-vs-deadline | Deterministically compare a deadline against the current instant; mandated path for temporal-awareness |
| current-tz | Read the operator's current IANA zone from tz_state, resolved by the host from live location; the single shared reader, no home-zone guessing |
| check-unanswered | Five-minute current-chat precheck; Jev screens dropped requests and wakes Claude only to answer confirmed drops |
Core rules keep binding constraints inline within 2,000 bytes each. The six trimmed rules link to skill-owned references containing only detail absent from the rule and owning SKILL.md; each clause has one home. Skills read those references for examples and context. Issue #129 defines the scope; Git history retains the original rules. tests/test_rule_references.py checks the byte limit, reference existence and owning skill links.
alwaysApply: true — the agent follows them without being prompted.nanoclaw-admin.ground-truth).See CHANGELOG.md for version history.
Renovate is this repo's only dependency scanner, configured in renovate.json. Its native pip requirements and GitHub Actions managers cover requirements-dev.txt and .github/workflows/. It runs in full mode, with regular updates scheduled before 06:00 UTC on Mondays and at most five concurrent regular-update PRs. When enabled, vulnerability alerts bypass that schedule and limit. Minor, patch and digest updates share one PR; major updates require Dependency Dashboard approval. Dependency commits use chore(deps).
Validate configuration changes with npx --yes --package renovate -- renovate-config-validator renovate.json.
tessl.json declares this repo's dev-time plugin dependencies.
jbaruch/* dependency floats at latest (Runtime-Managed Manifest Carve-Out, jbaruch/coding-policy: dependency-management).finsi/codex-review is third-party and pins. No dependency scanner covers the tessl ecosystem. Renewal cadence: quarterly — run tessl outdated and bump the pin in its own commit.