Agent skill system for authoring custom platform scripts#75
Merged
Conversation
Add AGENTS.md skeleton, five SKILL.md stubs under .agents/skills/, the probing evidence JSON Schema, the agent-reference corpus (samples-index, strategy-decision-tree, vendor-doc-search-recipes, empty failure-patterns), the samples-index build script, the agent-link validity check, and the two CI jobs that enforce them. Routing table, workflow algorithms, and SKILL.md bodies land in later phases.
…eed account as test, validate before import
…long-running shells, serialized-session pattern, real failure rows)
…eat, Get-Help discipline, probe shell hygiene, task-log shape)
…api pattern with auth-shape + one/two-step sub-dimensions
…wHash, add catch-block masking rule
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds a structured skill system that guides an AI agent through authoring,
validating, importing, triggering, and debugging Safeguard custom platform
scripts end-to-end.
What's in
AGENTS.mdorchestrator: operating modes (author-only / probe-only /full-loop), workflows for new and enhance, the iterative debug loop with
loop budget, and a routing table to the skills.
.agents/skills/:target-probing— learn how a live target behaves before authoring(paramiko-based SSH probing, probe-safety contract).
strategy-selection— pick a pattern, credential intent, and auth shapefrom probe evidence and vendor docs.
script-authoring— four pattern recipes (ssh-interactive, ssh-batch,http-api, http-form-fill) with a mandatory local-schema-validate inner
loop before any appliance round-trip.
safeguard-ps-operations— drive the appliance viasafeguard-ps(DeviceCode auth, validate / import / export / trigger / task-log fetch),
wrapping
tools/Invoke-PlatformDevLoop.ps1.task-log-analysis— classify failures and recommend the next iteration,backed by a failure-pattern catalog grown only from real runs.
docs/agent-reference/: samples index, strategydecision tree, failure patterns (seeded from real failures), vendor-doc
search recipes.
tools/Invoke-PlatformDevLoop.ps1,tools/Build-SamplesIndex.ps1,tools/Test-AgentLinks.ps1, and anevidence schema at
.agents/schemas/evidence.schema.json.Scope of testing
End-to-end validation so far has been SSH only — a full new-platform
workflow against a live SSH target, exercising probing, authoring, the
debug loop, and task-log analysis. HTTP recipes (
http-api,http-form-fill)are documented but have not yet been driven through an end-to-end loop;
they will be exercised in a follow-up.
No changes to existing samples, schema, or human-facing docs under
docs/concepts|guides|reference|tutorials|quick-start/.