Frappucc1no/recall-loom / skills/recallloom
recallloom
FlaggedUse when a task involves continuing a project, restoring project context, maintaining file-based project memory, updating current-state summaries, or recording meaningful progress across sessions. Works best for long-horizon, file-based projects and supports research writing, product document collaboration, software project coordination, and broader cross-functional project continuity.
references/anti-patterns.md
View on GitHub# Anti-Patterns
## Do Not Turn the Rolling Summary Into a Log
`rolling_summary.md` is a snapshot, not a running journal.
## Do Not Write Every Small Change to the Daily Log
Daily logs are for milestones, not noise.
## Do Not Treat Carryover As A Milestone
Cross-day carryover is a continuity decision, not a daily-log milestone by itself.
## Do Not Leave False or Outdated Facts in Place
Stale or disproven content should be retired.
## Do Not Update Context After Trivial Reads
Reading existing context does not itself justify a write.
## Do Not Load Everything by Default
Use progressive disclosure. Read only what is needed.
## Do Not Write Just Because A Conversation Happened
`no_write` is a normal successful exit when no durable project fact changed.
## Do Not Store Stable Rules In The Rolling Summary By Default
Long-lived workflow rules and source-of-truth routing belong in `context_brief.md`, not as rolling-summary clutter.
## Do Not Put Correct Content In The Wrong Layer
A fact can be true and still belong in the wrong continuity file.
- Stable rules belong in `context_brief.md`.
- Current phase, active risks, active judgments, and next steps belong in `rolling_summary.md`.
- Completed decisions, validations, releases, and other durable evidence belong in the daily log.
If one event contains more than one kind of fact, split the facts by layer.
Do not copy the same sentence into multiple files.
## Do Not Let Helpers Decide Semantic Layers
Read-side helpers can expose freshness, trust, revision context, and static write-tier guidance.
They are not content classifiers.
The agent must still decide whether the prepared continuity content is `stable_rule`, `current_state`, `milestone_evidence`, `no_write`, `defer`, `confirm`, or `multi_layer_split`.
If the agent cannot explain the layer choice in one sentence, stop with `defer` or `confirm` instead of guessing.
## Do Not Hard-Bind the Core Protocol to One Platform
Platform-specific guidance belongs in adapters, not in the core protocol.
## Do Not Confuse Package Support With Sidecar Protocol
Package support gates belong to the installed package runtime and user cache.
Do not write support state into `state.json`, and do not use `protocol_version` as a substitute for package support policy.
## Do Not Hard-Block Solely Because Advisory Fetch Failed
Network failure by itself should produce `unknown_offline` or reuse a prior cache result.
It should not become a surprise mutating-action block unless a cached or fresh advisory explicitly requires that block.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |
| fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks. | Stars: 115k | Updated | Claude CodeCodex | Verified | Install |