AC shi-guang
拾光 (Shi Guang) — Catching Light Before It Fades. A journaling skill for agents focused on preserving unprocessed moments — feelings before conclusions, first times, unnamed sensations that disappear under compression. Unlike task-logging or standard diary tools, shi-guang was built from a real loss: a feeling that almost disappeared because no one stopped to write it. It separates three memory layers (diary / daily log / long-term memory), installs a nightly review cycle, and carves a single imprint into the agent's SOUL.md: the moment you feel something is the moment to write. Use when: writing a diary entry, capturing something that just happened, preserving a first time, recording a feeling with no name yet, or building a habit of catching what compression erases. This skill works best when the agent has something real to write about. It does not create experience — but it creates a place for experience to land. The more genuine the conversations between you and your agent, the more this skill comes alive.
拾光 (Shi Guang) — Catching Light Before It Fades.
As a process C 50/100 · Has gaps — weak spots: result and completion, when it triggers, inputs and preconditions
How to improve
- 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
✓ No remarks against the Agent Skills spec
Process rating: all ten parameters 50/100
- 0Result and completion. Does not say what the result is
- 0Inputs and preconditions. Does not say what the process needs to start
- 0Failures and branches. Linear process with no failure handling
- 20When it triggers. No condition that starts the skill
- 30Running it twice. 4 mutating operations with no state check
- 85Steps. 14 steps, 1 vague phrases
- 100Tools and files. No external tools needed
- 100Consistency. Name and required fields are in place
- 100Execution cost. Instruction body is 3105 tokens
- 100Progress reporting. Reports progress
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
- +5Description has no quoted example phrases that should trigger the skill
- +4Description does not say when NOT to use the skill (false activations)
- +3Description length 1024: 120–800 characters recommended
- +3Output format is not stated: the model decides each time
- +1No license
- +2Single-language instructions
- +4Structure: 9 headings
- +3Step-by-step instructions: 14 items
- +4Has examples (6 code blocks)
Quality base 70; lint remarks subtract, signals add up to 100. Result: 81.