CB memory-review
Lightweight end-of-session memory review — scan the conversation just had for durable lessons and save them into Ralphy's tiered memory (`ralphy memory note`, write-and-tell: every save surfaced in chat, "forget <slug>" honored instantly), so the next session starts already knowing. The chat-native analog of hermes-agent's background self-review: signals are user corrections, frustration ("stop doing X", "I told you already"), durable preferences, model/provider facts discovered the hard way, and non-trivial fixes a future session would re-derive. Lighter than /postmortem — no 7-file set, no LLM call; the agent reads its own conversation and writes 0-5 entries. USE WHEN the user types "/memory-review", asks "what should we remember from this", "save the lessons", "update your memory", or a work session is wrapping up after >=1 user correction / re-roll / model swap (fire proactively then). DO NOT FIRE when a full postmortem is warranted (>=2 corrections or >=1 CLI gap — that is /postmortem, which ends with `ralphy memory distill`); for capturing ONE remark mid-flight (that is invariant #18 — note it immediately, no review pass needed); for dev-mode sessions on the Ralphy repo itself (lessons go to notes/, not memory). See body for the signal list, the do-not-capture list, and HARD INVARIANTS.
Lightweight end-of-session memory review — scan the conversation just had for durable lessons and save them into Ralphy's tiered memory (ralphy memory note…
As a process B 69/100 · Nearly there — weak spots: result and completion, inputs and preconditions, progress reporting
How to improve
- Shorten the description to 1024 characters.
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 0
✓ No critical or high findings
Files scanned: 1. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- error
description-longdescription is 1313 chars, limit 1024 - note
frontmatter-keyunknown frontmatter key "namespace"
Process rating: all ten parameters 69/100
- 0Result and completion. Does not say what the result is
- 0Inputs and preconditions. Does not say what the process needs to start
- 0Progress reporting. Says nothing while it works
- 70When it triggers. States when to use, but not when not to
- 100Tools and files. No external tools needed
- 100Steps. 23 steps
- 100Failures and branches. 1 branches, has a failure section
- 100Consistency. Name and required fields are in place
- 100Execution cost. Instruction body is 1011 tokens
- 100Running it twice. No mutating operations
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- low The response is described with custom markup (4 tags): a typed call is more reliable
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +4Description does not say when NOT to use the skill (false activations)
- +3Description length 1313: 120–800 characters recommended
- +3Output format is not stated: the model decides each time
- +4No input/output examples
- +1No license
- +2Single-language instructions
- +5Description quotes 7 example trigger phrases
- +4Structure: 4 headings
- +3Step-by-step instructions: 23 items
Quality base 70; lint remarks subtract, signals add up to 100. Result: 56.