mirror of
https://github.com/magnus919/agent-skills.git
synced 2026-09-15 21:46:29 +03:00
Magnus Hedemarkandfactory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
2c78dfcca3
Add the semantic-spacetime skill skeleton: a thin SKILL.md router with triggers/anti-triggers, Load By Need, Quick Start, Related Skills, gotchas, and exit conditions; a human-facing README; MIT license; deep provenance-marked theory references (foundations, glossary, bibliography); the versioned sst-model-v1 template pair; a 6-case schema-v1 eval manifest; the root README catalog entry; regenerated marketplace and llms artifacts; and the skill-triggers row. Co-authored-by: factory-droid[bot] <138933559+factory-droid[bot]@users.noreply.github.com>
87 lines
3.5 KiB
Cheetah
87 lines
3.5 KiB
Cheetah
# Semantic Spacetime Analysis: <analysis-title>
|
|
|
|
Semantic Spacetime analysis report. Fill this skeleton from a completed model
|
|
in [sst-model.yaml.tmpl](sst-model.yaml.tmpl) (the versioned `sst-model-v1`
|
|
format). Every section below corresponds to a construct defined in that model
|
|
— agents, nodes, edges, acceptances, trajectories, observations — so the
|
|
report and the model stay in sync; do not invent new fields.
|
|
|
|
- Analysis version: <0.1.0>
|
|
- Status: <DRAFTED | IN_REVIEW | COMPLETE>
|
|
- Date: <YYYY-MM-DD>
|
|
- Model version analyzed: <sst-model-v1>
|
|
- Analyst: <agent-id or human name>
|
|
|
|
## 1. System description
|
|
|
|
Describe the system being modeled in plain language: who the agents are (their
|
|
ids and roles from the model's `agents`), what the semantic spacetime covers,
|
|
and the question the analysis answers (e.g., "why did the two agents'
|
|
interpretations diverge?").
|
|
|
|
- Agents: <agent-ids and roles>
|
|
- Scope of the spacetime: <what the nodes/edges cover>
|
|
- Analysis question: <the question this report answers>
|
|
|
|
## 2. Semantic spacetime map
|
|
|
|
Summarize the model's graph as built in `sst-model.yaml.tmpl`:
|
|
|
|
- Nodes: <node-ids grouped by type: event | thing | concept>
|
|
- Edges: <the gamma(3,4) edge set, each as from --link-value--> to, with the
|
|
link meaning 0 = NEAR, +/-1 = LEADS TO, +/-2 = CONTAINS, +/-3 = EXPRESSES>
|
|
- Acceptances: <promise -> accepting agent pairs from the model's `acceptances`>
|
|
- Trajectories: <the declared `trajectories`, i.e., the paths intent takes
|
|
through the graph>
|
|
- Observations: <the proper-time record: what changed and when, from the
|
|
model's `observations`>
|
|
|
|
## 3. Findings: drift, divergence, absorbing states
|
|
|
|
List each finding with the evidence that grounds it (which nodes, edges,
|
|
trajectories, or observations support it):
|
|
|
|
### Finding <F1> — <short name>
|
|
- Observation: <what the record shows — quote the relevant `observations` entries>
|
|
- Interpretation: <which semantic elements drifted apart, which trajectory
|
|
diverged, or which node is an absorbing state>
|
|
- Absorbing-state / information-leak check: <if a partial-graph dead-end is
|
|
present, state where information stops flowing and where intent must be
|
|
injected>
|
|
|
|
### Finding <F2> — <short name>
|
|
- Observation: <...>
|
|
- Interpretation: <...>
|
|
- Absorbing-state / information-leak check: <...>
|
|
|
|
## 4. Interventions
|
|
|
|
For each finding, propose a concrete change to the model (and, if applicable,
|
|
to the system it models):
|
|
|
|
- <F1> intervention: <e.g., re-anchor a dangling concept to a thing or event;
|
|
add an edge to close the leak; renegotiate a promise whose acceptance is
|
|
missing>
|
|
- <F2> intervention: <e.g., record a new observation at a finer timescale
|
|
(timelike measurement) before asserting the semantics>
|
|
|
|
## 5. Verification / measurement plan
|
|
|
|
State how each intervention will be measured, using the model's own record
|
|
machinery:
|
|
|
|
- What to observe: <new `observations` entries or `trajectories` to add>
|
|
- Measurement mode: <spacelike (repeated trials, constant state) | timelike
|
|
(continuous accumulation) — see the skill's foundations reference>
|
|
- Success criterion: <the observable that must hold, e.g., semantic distance
|
|
between the two `concept` nodes stays at 0 NEAR across N observations>
|
|
- Refresh budget: <how often the interpretation must be re-confirmed — stale
|
|
knowledge decays geometrically unless refreshed>
|
|
|
|
## 6. Exit
|
|
|
|
This analysis is complete when the system is modeled as a semantic spacetime,
|
|
findings are recorded with evidence, and the verification plan is stated.
|
|
When diagnosing drift: stop after three non-converging passes and report the
|
|
evidence.
|