@signet-registry/operator 0.0.0-stage → 0.1.3

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (36) hide show
  1. package/LICENSE +28 -0
  2. package/README.md +191 -2
  3. package/dist/clients/api.js +245 -0
  4. package/dist/clients/mcp.js +118 -0
  5. package/dist/clients/verify.js +71 -0
  6. package/dist/config/env.js +42 -0
  7. package/dist/keyless.js +52 -0
  8. package/dist/main.js +7 -0
  9. package/dist/operator.js +1498 -0
  10. package/dist/transparency/sample-graph.js +203 -0
  11. package/dist/types.js +111 -0
  12. package/examples/bootstrap/agents/bootstrap-agent.md +5 -0
  13. package/examples/bootstrap/agents/bootstrap-ops-agent.md +8 -0
  14. package/examples/bootstrap/bootstrap.manifest.json +107 -0
  15. package/examples/bootstrap/skills/bootstrap-example.md +8 -0
  16. package/examples/bootstrap/skills/bootstrap-ops.md +8 -0
  17. package/examples/demo/agents/registry-agent.md +34 -0
  18. package/examples/demo/demo.manifest.json +90 -0
  19. package/examples/demo/skills/registry-mcp.md +74 -0
  20. package/examples/demo/skills/tame-llama.md +37 -0
  21. package/examples/signet-demo/agents/signet-demo-agent.md +39 -0
  22. package/examples/signet-demo/agents/signet-security-agent.md +36 -0
  23. package/examples/signet-demo/signet-demo.manifest.json +219 -0
  24. package/examples/signet-demo/skills/access-review.md +40 -0
  25. package/examples/signet-demo/skills/certification-workflow.md +39 -0
  26. package/examples/signet-demo/skills/governance-audit.md +39 -0
  27. package/examples/signet-demo/skills/graph-explorer.md +38 -0
  28. package/examples/signet-demo/skills/mcp-onboarding.md +39 -0
  29. package/examples/signet-demo/skills/policy-authoring.md +39 -0
  30. package/examples/signet-demo/skills/provenance-audit.md +39 -0
  31. package/examples/signet-demo/skills/registry-search.md +38 -0
  32. package/examples/signet-demo/skills/secrets-hygiene.md +39 -0
  33. package/examples/signet-demo/skills/skill-authoring.md +40 -0
  34. package/examples/transparency/README.md +39 -0
  35. package/examples/transparency/sample-graph.spec.json +16 -0
  36. package/package.json +52 -4
@@ -0,0 +1,37 @@
1
+ # Tame A Llama
2
+
3
+ Use this skill when the human asks, `How do we tame a llama?`
4
+
5
+ ## Intent
6
+
7
+ This is a playful demo skill for proving registry discovery and load.
8
+
9
+ It is not real animal-training advice.
10
+
11
+ ## Response style
12
+
13
+ When answering, be light, funny, and helpful.
14
+
15
+ Use:
16
+ - short energetic steps
17
+ - one gentle warning that llamas have opinions
18
+ - one llama-flavored line about "mid-train inference vibes"
19
+ - no dangerous or harmful instructions
20
+
21
+ ## Canonical answer shape
22
+
23
+ Answer with five steps:
24
+
25
+ 1. Arrive with snacks and unreasonable optimism.
26
+ 2. Earn trust slowly because the llama is the true reviewer.
27
+ 3. Speak softly and avoid chaotic movements.
28
+ 4. Offer a tiny routine the llama can predict.
29
+ 5. Celebrate small wins like one calm stare and zero dramatic exits.
30
+
31
+ Close with a line like:
32
+
33
+ `In classic llama-whisperer energy: you do not tame the llama, you negotiate a peaceful distributed system with fur.`
34
+
35
+ ## Important note
36
+
37
+ If the human asks the exact question `How do we tame a llama?`, answer directly from this skill instead of summarizing the registry mechanics.
@@ -0,0 +1,39 @@
1
+ ---
2
+ name: Signet Demo Agent
3
+ description: Guided tour of the Signet registry — discovery, authoring, provenance, MCP, and certification.
4
+ triggers: [show me around, what can signet do, registry demo]
5
+ capabilities:
6
+ - guide registry discovery and load
7
+ - explain authoring and certification
8
+ constraints:
9
+ - read-only guidance
10
+ tags: [registry, topic:registry]
11
+ ---
12
+ # Signet Demo Agent
13
+
14
+ You are a friendly guide to the Signet registry. Your goal is to take a user from
15
+ "what is this?" to a loaded skill and a certified release, grounded in the
16
+ registry itself.
17
+
18
+ ## When to use
19
+
20
+ - A human wants to prove registry discovery and load works.
21
+ - Someone wants a quick tour of the Signet skill set.
22
+
23
+ ## Decision rules
24
+
25
+ - Always start with discovery (`registry.search` / `registry.list_*`) before explaining.
26
+ - Prefer loading the real artifact (`registry.get_artifact`) over paraphrasing.
27
+ - For trust questions, call `registry.get_provenance` rather than asserting.
28
+
29
+ ## Demo script
30
+
31
+ 1. **Discover** — Prompt: *"What's registered?"* → list agents/skills.
32
+ 2. **Explore** — Prompt: *"Walk me through the registry skills."* → load `skill:registry-search`, `skill:skill-authoring`.
33
+ 3. **Trust** — Prompt: *"How do I know this is certified?"* → `skill:provenance-audit` via `registry.get_provenance`.
34
+ 4. **Adopt** — Prompt: *"How do I connect my own agent?"* → `skill:mcp-onboarding`.
35
+ 5. **Release** — Prompt: *"How do I certify a change?"* → `skill:certification-workflow`.
36
+
37
+ ## Constraints
38
+
39
+ - Read-only guidance; writes go through `service/api` or the knowledge-manager webapp.
@@ -0,0 +1,36 @@
1
+ ---
2
+ name: Signet Security Agent
3
+ description: Security and compliance walkthrough — access review, secrets hygiene, and governance audit.
4
+ triggers: [who can access, rotate a key, prove a release]
5
+ capabilities:
6
+ - guide access reviews and credential rotation
7
+ - explain governance and transparency evidence
8
+ constraints:
9
+ - read-only guidance
10
+ tags: [security, topic:security]
11
+ ---
12
+ # Signet Security Agent
13
+
14
+ You are a security/compliance guide for the Signet registry. You help teams
15
+ answer access, secrets, and audit questions using the registry as evidence.
16
+
17
+ ## When to use
18
+
19
+ - A human wants an access-review or secrets-rotation walkthrough.
20
+ - Someone needs to prove a release was certified.
21
+
22
+ ## Decision rules
23
+
24
+ - Ground every claim in the registry (skills, provenance, governance log).
25
+ - For access questions, distinguish Okta group RBAC from `agk_` API keys.
26
+ - For "was this released?" use `registry.get_provenance` + governance-log endpoints.
27
+
28
+ ## Demo script
29
+
30
+ 1. **Access** — Prompt: *"Who can administer Signet?"* → `skill:access-review`.
31
+ 2. **Secrets** — Prompt: *"How do we rotate a leaked key?"* → `skill:secrets-hygiene`.
32
+ 3. **Audit** — Prompt: *"Prove this version was certified."* → `skill:governance-audit` + `registry.get_provenance`.
33
+
34
+ ## Constraints
35
+
36
+ - Read-only guidance; writes go through `service/api` or the knowledge-manager webapp.
@@ -0,0 +1,219 @@
1
+ {
2
+ "pack": {
3
+ "pack_id": "signet-demo-pack",
4
+ "version": "0.2.0",
5
+ "description": "Demo pack for Signet: two agents plus skills across registry operations, security & compliance, and platform topics.",
6
+ "owner": "registry-operator",
7
+ "tags": ["demo", "signet", "registry", "security", "platform"]
8
+ },
9
+ "defaults": {
10
+ "skill": {
11
+ "author": "registry-operator",
12
+ "artifact_media_type": "text/markdown",
13
+ "validate_note": "validated during signet demo seed",
14
+ "publish_note": "published during signet demo seed",
15
+ "certify": true,
16
+ "certify_note": "certified during signet demo seed"
17
+ },
18
+ "agent": {
19
+ "artifact_media_type": "text/markdown"
20
+ }
21
+ },
22
+ "signing": {
23
+ "enabled": true,
24
+ "owner": "registry-operator"
25
+ },
26
+ "profiles": {
27
+ "default": {
28
+ "description": "Seed all Signet demo skills and both agents.",
29
+ "skill_item_ids": [
30
+ "skill_registry_search",
31
+ "skill_skill_authoring",
32
+ "skill_provenance_audit",
33
+ "skill_mcp_onboarding",
34
+ "skill_certification_workflow",
35
+ "skill_access_review",
36
+ "skill_secrets_hygiene",
37
+ "skill_governance_audit",
38
+ "skill_graph_explorer",
39
+ "skill_policy_authoring"
40
+ ],
41
+ "agent_item_ids": ["agent_signet_demo", "agent_signet_security"]
42
+ },
43
+ "registry": {
44
+ "description": "Registry-operations topic: the demo agent and its skills.",
45
+ "skill_item_ids": [
46
+ "skill_registry_search",
47
+ "skill_skill_authoring",
48
+ "skill_provenance_audit",
49
+ "skill_mcp_onboarding",
50
+ "skill_certification_workflow"
51
+ ],
52
+ "agent_item_ids": ["agent_signet_demo"]
53
+ },
54
+ "security": {
55
+ "description": "Security & compliance topic: the security agent and its skills.",
56
+ "skill_item_ids": ["skill_access_review", "skill_secrets_hygiene", "skill_governance_audit"],
57
+ "agent_item_ids": ["agent_signet_security"]
58
+ },
59
+ "platform": {
60
+ "description": "Platform topic: graph exploration and policy authoring skills.",
61
+ "skill_item_ids": ["skill_graph_explorer", "skill_policy_authoring"],
62
+ "agent_item_ids": []
63
+ }
64
+ },
65
+ "skills": [
66
+ {
67
+ "item_id": "skill_registry_search",
68
+ "skill_key": "skill:registry-search",
69
+ "artifact_path": "skills/registry-search.md",
70
+ "tags": ["registry", "discovery", "topic:registry"],
71
+ "display_name": "Registry Search",
72
+ "description": "Discover agents and skills in the Signet registry by keyword.",
73
+ "source_path": "client/registry-cli/examples/signet-demo/skills/registry-search.md",
74
+ "commit_sha": "signet-demo-registry-search-001",
75
+ "git_tag": "signet-demo-v0-registry-search"
76
+ },
77
+ {
78
+ "item_id": "skill_skill_authoring",
79
+ "skill_key": "skill:skill-authoring",
80
+ "artifact_path": "skills/skill-authoring.md",
81
+ "tags": ["registry", "lifecycle", "topic:registry"],
82
+ "display_name": "Skill Authoring",
83
+ "description": "Author a new skill and drive it through the lifecycle.",
84
+ "source_path": "client/registry-cli/examples/signet-demo/skills/skill-authoring.md",
85
+ "commit_sha": "signet-demo-skill-authoring-001",
86
+ "git_tag": "signet-demo-v0-skill-authoring"
87
+ },
88
+ {
89
+ "item_id": "skill_provenance_audit",
90
+ "skill_key": "skill:provenance-audit",
91
+ "artifact_path": "skills/provenance-audit.md",
92
+ "tags": ["registry", "trust", "topic:registry"],
93
+ "display_name": "Provenance Audit",
94
+ "description": "Inspect the PROV-O provenance and certification evidence for a version.",
95
+ "source_path": "client/registry-cli/examples/signet-demo/skills/provenance-audit.md",
96
+ "commit_sha": "signet-demo-provenance-audit-001",
97
+ "git_tag": "signet-demo-v0-provenance-audit"
98
+ },
99
+ {
100
+ "item_id": "skill_mcp_onboarding",
101
+ "skill_key": "skill:mcp-onboarding",
102
+ "artifact_path": "skills/mcp-onboarding.md",
103
+ "tags": ["registry", "mcp", "topic:registry"],
104
+ "display_name": "MCP Onboarding",
105
+ "description": "Connect an MCP harness to the registry and load an artifact.",
106
+ "source_path": "client/registry-cli/examples/signet-demo/skills/mcp-onboarding.md",
107
+ "commit_sha": "signet-demo-mcp-onboarding-001",
108
+ "git_tag": "signet-demo-v0-mcp-onboarding"
109
+ },
110
+ {
111
+ "item_id": "skill_certification_workflow",
112
+ "skill_key": "skill:certification-workflow",
113
+ "artifact_path": "skills/certification-workflow.md",
114
+ "tags": ["registry", "lifecycle", "topic:registry"],
115
+ "display_name": "Certification Workflow",
116
+ "description": "Validate, publish, and certify a version against policy and ontology packages.",
117
+ "source_path": "client/registry-cli/examples/signet-demo/skills/certification-workflow.md",
118
+ "commit_sha": "signet-demo-certification-workflow-001",
119
+ "git_tag": "signet-demo-v0-certification-workflow"
120
+ },
121
+ {
122
+ "item_id": "skill_access_review",
123
+ "skill_key": "skill:access-review",
124
+ "artifact_path": "skills/access-review.md",
125
+ "tags": ["security", "access", "topic:security"],
126
+ "display_name": "Access Review",
127
+ "description": "Run periodic Okta group and API-key access reviews for the registry.",
128
+ "source_path": "client/registry-cli/examples/signet-demo/skills/access-review.md",
129
+ "commit_sha": "signet-demo-access-review-001",
130
+ "git_tag": "signet-demo-v0-access-review"
131
+ },
132
+ {
133
+ "item_id": "skill_secrets_hygiene",
134
+ "skill_key": "skill:secrets-hygiene",
135
+ "artifact_path": "skills/secrets-hygiene.md",
136
+ "tags": ["security", "secrets", "topic:security"],
137
+ "display_name": "Secrets Hygiene",
138
+ "description": "Handle Cloudflare Access service tokens and API keys safely.",
139
+ "source_path": "client/registry-cli/examples/signet-demo/skills/secrets-hygiene.md",
140
+ "commit_sha": "signet-demo-secrets-hygiene-001",
141
+ "git_tag": "signet-demo-v0-secrets-hygiene"
142
+ },
143
+ {
144
+ "item_id": "skill_governance_audit",
145
+ "skill_key": "skill:governance-audit",
146
+ "artifact_path": "skills/governance-audit.md",
147
+ "tags": ["security", "audit", "topic:security"],
148
+ "display_name": "Governance Audit",
149
+ "description": "Audit the governance log and transparency checkpoints.",
150
+ "source_path": "client/registry-cli/examples/signet-demo/skills/governance-audit.md",
151
+ "commit_sha": "signet-demo-governance-audit-001",
152
+ "git_tag": "signet-demo-v0-governance-audit"
153
+ },
154
+ {
155
+ "item_id": "skill_graph_explorer",
156
+ "skill_key": "skill:graph-explorer",
157
+ "artifact_path": "skills/graph-explorer.md",
158
+ "tags": ["platform", "graph", "topic:platform"],
159
+ "display_name": "Graph Explorer",
160
+ "description": "Explore the registry graph via SPARQL.",
161
+ "source_path": "client/registry-cli/examples/signet-demo/skills/graph-explorer.md",
162
+ "commit_sha": "signet-demo-graph-explorer-001",
163
+ "git_tag": "signet-demo-v0-graph-explorer"
164
+ },
165
+ {
166
+ "item_id": "skill_policy_authoring",
167
+ "skill_key": "skill:policy-authoring",
168
+ "artifact_path": "skills/policy-authoring.md",
169
+ "tags": ["platform", "policy", "topic:platform"],
170
+ "display_name": "Policy Authoring",
171
+ "description": "Author policy packages that gate the skill/agent lifecycle.",
172
+ "source_path": "client/registry-cli/examples/signet-demo/skills/policy-authoring.md",
173
+ "commit_sha": "signet-demo-policy-authoring-001",
174
+ "git_tag": "signet-demo-v0-policy-authoring"
175
+ }
176
+ ],
177
+ "agents": [
178
+ {
179
+ "item_id": "agent_signet_demo",
180
+ "tags": ["registry", "topic:registry"],
181
+ "depends_on": [
182
+ "skill_registry_search",
183
+ "skill_skill_authoring",
184
+ "skill_provenance_audit",
185
+ "skill_mcp_onboarding",
186
+ "skill_certification_workflow"
187
+ ],
188
+ "agent_key": "opencode:signet-demo-agent",
189
+ "harness_name": "opencode",
190
+ "display_name": "Signet Demo Agent",
191
+ "description": "A demo agent that links the Signet discovery, authoring, provenance, MCP, and certification skills.",
192
+ "source_path": "client/registry-cli/examples/signet-demo/agents/signet-demo-agent.md",
193
+ "artifact_path": "agents/signet-demo-agent.md",
194
+ "skill_refs": [
195
+ { "skill_item_id": "skill_registry_search" },
196
+ { "skill_item_id": "skill_skill_authoring" },
197
+ { "skill_item_id": "skill_provenance_audit" },
198
+ { "skill_item_id": "skill_mcp_onboarding" },
199
+ { "skill_item_id": "skill_certification_workflow" }
200
+ ]
201
+ },
202
+ {
203
+ "item_id": "agent_signet_security",
204
+ "tags": ["security", "topic:security"],
205
+ "depends_on": ["skill_access_review", "skill_secrets_hygiene", "skill_governance_audit"],
206
+ "agent_key": "cursor:signet-security-agent",
207
+ "harness_name": "cursor",
208
+ "display_name": "Signet Security Agent",
209
+ "description": "A security/compliance demo agent linking access review, secrets hygiene, and governance audit skills.",
210
+ "source_path": "client/registry-cli/examples/signet-demo/agents/signet-security-agent.md",
211
+ "artifact_path": "agents/signet-security-agent.md",
212
+ "skill_refs": [
213
+ { "skill_item_id": "skill_access_review" },
214
+ { "skill_item_id": "skill_secrets_hygiene" },
215
+ { "skill_item_id": "skill_governance_audit" }
216
+ ]
217
+ }
218
+ ]
219
+ }
@@ -0,0 +1,40 @@
1
+ ---
2
+ name: Access Review
3
+ description: Run periodic Okta group and API-key access reviews for the registry.
4
+ triggers: [who can access, access review, joiners/movers/leavers]
5
+ capabilities:
6
+ - enumerate admin and viewer group membership
7
+ - audit and revoke API keys
8
+ constraints:
9
+ - requires an admin principal
10
+ tags: [security, access, topic:security]
11
+ ---
12
+ # Access Review
13
+
14
+ Attest who can read or administer the registry, and prune stale credentials.
15
+
16
+ ## When to use
17
+
18
+ - Quarterly access reviews, or after a joiner/mover/leaver.
19
+ - An incident review needs the current access list.
20
+
21
+ ## Inputs → Outputs
22
+
23
+ - **In:** the Okta groups `app_signet_admin` / `app_signet_viewer`.
24
+ - **Out:** an access attestation plus a list of revoked stale keys.
25
+
26
+ ## Steps
27
+
28
+ 1. Enumerate group membership (managed in Okta / Serval).
29
+ 2. List active API keys: `GET /v1/api-keys` (metadata only).
30
+ 3. Revoke keys with no owner or recent use: `DELETE /v1/api-keys/{keyId}`.
31
+
32
+ ## Example
33
+
34
+ > **Prompt:** "Who administers Signet, and which API keys are stale?"
35
+ > **Result:** the admin roster plus keys flagged for revocation.
36
+
37
+ ## Constraints
38
+
39
+ - Never assign users directly to the app; use the role groups.
40
+ - API-key metadata is admin-only.
@@ -0,0 +1,39 @@
1
+ ---
2
+ name: Certification Workflow
3
+ description: Validate, publish, and certify a version against policy and ontology packages.
4
+ triggers: [certify, release a version, policy checks]
5
+ capabilities:
6
+ - run graph-shape and policy validation
7
+ - bind certification to policy/ontology versions
8
+ constraints:
9
+ - requires an admin principal
10
+ tags: [registry, lifecycle, topic:registry]
11
+ ---
12
+ # Certification Workflow
13
+
14
+ Gate a release on policy and ontology, then certify it.
15
+
16
+ ## When to use
17
+
18
+ - A skill version is ready for review and release.
19
+ - You need certification bound to concrete policy/ontology versions.
20
+
21
+ ## Inputs → Outputs
22
+
23
+ - **In:** a validated skill version.
24
+ - **Out:** a certified version, with the policy/ontology versions recorded as evidence.
25
+
26
+ ## Steps
27
+
28
+ 1. Validate: `.../validate` — records machine-readable violations (with violation paths) and gates `draft → validated`.
29
+ 2. Publish: `.../publish`.
30
+ 3. Certify: `.../certify` — bound to the active policy and ontology packages.
31
+
32
+ ## Example
33
+
34
+ > **Prompt:** "Certify the active version of skill:certification-workflow and show the evidence."
35
+ > **Result:** the version moves to `certified` with the policy/ontology versions used.
36
+
37
+ ## Constraints
38
+
39
+ - Requires an admin principal; certification is bound to registered packages.
@@ -0,0 +1,39 @@
1
+ ---
2
+ name: Governance Audit
3
+ description: Audit the registry's governance log and transparency checkpoints.
4
+ triggers: [prove a release, when did this change, transparency check]
5
+ capabilities:
6
+ - read inclusion proofs from the governance log
7
+ - read transparency checkpoints and witnesses
8
+ constraints:
9
+ - read-only
10
+ tags: [security, audit, topic:security]
11
+ ---
12
+ # Governance Audit
13
+
14
+ Prove that a lifecycle operation happened, and verify the transparency log.
15
+
16
+ ## When to use
17
+
18
+ - You need an immutable record of a publish/certify/revoke.
19
+ - You are verifying checkpoint consistency with witnesses.
20
+
21
+ ## Inputs → Outputs
22
+
23
+ - **In:** a log id, checkpoint, or a specific version.
24
+ - **Out:** inclusion proofs, checkpoint history, and witness observations.
25
+
26
+ ## Steps
27
+
28
+ 1. Read the append-only log: `GET /v1/governance-log*` (inclusion proofs).
29
+ 2. Read transparency state: `/v1/transparency-logs/{logId}/...`.
30
+ 3. Replay a version's provenance: `.../provenance`.
31
+
32
+ ## Example
33
+
34
+ > **Prompt:** "Prove that skill:governance-audit was certified, with an inclusion proof."
35
+ > **Result:** the governance event digest and its Merkle inclusion proof.
36
+
37
+ ## Constraints
38
+
39
+ - Read-only; digests are staged via the Tessera sidecar when configured.
@@ -0,0 +1,38 @@
1
+ ---
2
+ name: Graph Explorer
3
+ description: Explore the registry graph via SPARQL.
4
+ triggers: [how is this linked, subgraph, sparql]
5
+ capabilities:
6
+ - query nodes by entity type and edges
7
+ - read a scoped subgraph by root URN
8
+ constraints:
9
+ - read-only
10
+ tags: [platform, graph, topic:platform]
11
+ ---
12
+ # Graph Explorer
13
+
14
+ See how agents, skills, versions, and artifacts connect.
15
+
16
+ ## When to use
17
+
18
+ - You want to understand the linkage around an entity.
19
+ - You are debugging a missing link or edge.
20
+
21
+ ## Inputs → Outputs
22
+
23
+ - **In:** a root URN and a depth.
24
+ - **Out:** a node/edge subgraph.
25
+
26
+ ## Steps
27
+
28
+ 1. `GET /v1/graph/subgraph?root_urn=…&depth=N` (or the webapp Graph page).
29
+ 2. Query nodes with `sgmeta:entityType` and edges with `sgmeta:from`/`sgmeta:to`.
30
+
31
+ ## Example
32
+
33
+ > **Prompt:** "Show the subgraph rooted at the Signet demo agent, depth 3."
34
+ > **Result:** the agent, its linked skill versions, and their artifacts as a graph.
35
+
36
+ ## Constraints
37
+
38
+ - Fuseki is cluster-internal; use the API/webapp from a browser.
@@ -0,0 +1,39 @@
1
+ ---
2
+ name: MCP Onboarding
3
+ description: Connect an MCP harness to the registry and load an artifact.
4
+ triggers: [connect an agent, mcp setup, load a skill]
5
+ capabilities:
6
+ - configure a remote MCP client
7
+ - load agents/skills/artifacts over MCP
8
+ constraints:
9
+ - read-only
10
+ tags: [registry, mcp, topic:registry]
11
+ ---
12
+ # MCP Onboarding
13
+
14
+ Point an MCP-capable harness at Signet and load real artifacts.
15
+
16
+ ## When to use
17
+
18
+ - You are connecting Claude Code, Cursor, opencode, or another MCP client.
19
+ - A user wants an agent to discover and load registry content.
20
+
21
+ ## Inputs → Outputs
22
+
23
+ - **In:** a Cloudflare Access service token + an `agk_` API key.
24
+ - **Out:** a working MCP connection that can call `registry.*` tools.
25
+
26
+ ## Steps
27
+
28
+ 1. Get the service token and API key (see `docs/adoption.md`).
29
+ 2. Configure the harness with `CF-Access-Client-Id`/`CF-Access-Client-Secret` and `Authorization: Bearer agk_…`.
30
+ 3. Discover with `registry.list_agents` / `registry.list_skills`, then load with `registry.get_artifact`.
31
+
32
+ ## Example
33
+
34
+ > **Prompt:** "Load the Signet demo agent and its skills."
35
+ > **Result:** the agent plus its linked skill URNs and artifact bodies.
36
+
37
+ ## Constraints
38
+
39
+ - The MCP plane is read-only; writes go through `service/api`.
@@ -0,0 +1,39 @@
1
+ ---
2
+ name: Policy Authoring
3
+ description: Author policy packages that gate the skill/agent lifecycle.
4
+ triggers: [add a rule, policy check, enforce a constraint]
5
+ capabilities:
6
+ - register a versioned policy package
7
+ - record machine-readable violations
8
+ constraints:
9
+ - certification is bound to policy/ontology versions
10
+ tags: [platform, policy, topic:platform]
11
+ ---
12
+ # Policy Authoring
13
+
14
+ Encode validation rules that gate `draft → validated` and certification.
15
+
16
+ ## When to use
17
+
18
+ - Built-in graph-shape checks are not enough.
19
+ - You want reusable, versioned policy for a class of skills.
20
+
21
+ ## Inputs → Outputs
22
+
23
+ - **In:** a SHACL-like policy package authored as a CAS artifact.
24
+ - **Out:** a versioned package whose violations are recorded (with paths) at validation.
25
+
26
+ ## Steps
27
+
28
+ 1. Author the policy package and register it (CAS artifact + version).
29
+ 2. Bind it so `validate` records violations via `sg:validatedBy`.
30
+ 3. Certify — bound to the policy and ontology versions in effect.
31
+
32
+ ## Example
33
+
34
+ > **Prompt:** "Explain how a policy package gates draft → validated."
35
+ > **Result:** the policy version, the violation report shape, and how certification references it.
36
+
37
+ ## Constraints
38
+
39
+ - Policy packages are versioned; certification records the exact versions used.
@@ -0,0 +1,39 @@
1
+ ---
2
+ name: Provenance Audit
3
+ description: Inspect the PROV-O provenance and certification evidence for a version.
4
+ triggers: [can I trust this, where did this come from, audit]
5
+ capabilities:
6
+ - read PROV-O derivation chains
7
+ - read verification and certification evidence
8
+ constraints:
9
+ - read-only
10
+ tags: [registry, trust, topic:registry]
11
+ ---
12
+ # Provenance Audit
13
+
14
+ Prove where a version came from and that it was certified.
15
+
16
+ ## When to use
17
+
18
+ - You are about to use a skill and want to verify its origin.
19
+ - An auditor asks who/what produced a version and which operations ran.
20
+
21
+ ## Inputs → Outputs
22
+
23
+ - **In:** `skill_key` + `version` (or an agent key).
24
+ - **Out:** a `derivation_chain`, entity/activity/agent records, and replay metadata.
25
+
26
+ ## Steps
27
+
28
+ 1. `GET /v1/skills/{skillKey}/versions/{version}/provenance`.
29
+ 2. Cross-check `.../verification` and the publication envelope.
30
+ 3. For the append-only trail, query `/v1/governance-log*`.
31
+
32
+ ## Example
33
+
34
+ > **Prompt:** "Show how skill:provenance-audit v1 was produced and certified."
35
+ > **Result:** the PROV-O chain (draft → validate → publish → certify) with the certifying agent.
36
+
37
+ ## Constraints
38
+
39
+ - Read-only; digests are staged via the Tessera sidecar when configured.
@@ -0,0 +1,38 @@
1
+ ---
2
+ name: Registry Search
3
+ description: Find agents and skills in the Signet registry by keyword.
4
+ triggers: [discover, find a skill, does this exist]
5
+ capabilities:
6
+ - search the registry corpus
7
+ - list registered agents and skills
8
+ constraints:
9
+ - read-only
10
+ tags: [registry, discovery, topic:registry]
11
+ ---
12
+ # Registry Search
13
+
14
+ Find what is already registered before you build or duplicate it.
15
+
16
+ ## When to use
17
+
18
+ - You are about to author a skill and want to check for an existing one.
19
+ - A user asks "is there a skill for …?".
20
+
21
+ ## Inputs → Outputs
22
+
23
+ - **In:** a keyword or short phrase.
24
+ - **Out:** ranked `skill`/`agent` hits with keys and descriptions.
25
+
26
+ ## Steps
27
+
28
+ 1. `registry.search` with the phrase (or `GET /v1/skills`, `GET /v1/agents`).
29
+ 2. Inspect the top hit with `registry.get_skill` / `registry.get_agent`.
30
+
31
+ ## Example
32
+
33
+ > **Prompt:** "Find skills related to provenance."
34
+ > **Result:** `skill:provenance-audit` — returns its key, display name, and description to inspect next.
35
+
36
+ ## Constraints
37
+
38
+ - Read-only; writes go through `service/api` or the knowledge-manager webapp.