docs(skills): add missing skill stubs, interactive patterns wizard, d… by mangod12 · Pull Request #5111 · aden-hive/hive

mangod12 · 2026-02-19T10:53:01Z

…ebugger fix applier, test preview

Fixes several workflow cross-compatibility gaps:

Add missing hive-debugger workflow stub (.agent/workflows/hive-debugger.md)
Add missing triage-issue workflow stub (.agent/workflows/triage-issue.md)
Add missing hive-debugger and triage-issue symlinks in .agent/skills/ and .agents/skills/
Enrich all 6 existing workflow stubs with description and when_to_use metadata
hive.md: slim from ~500 to 151 lines as pure routing document (remove duplicate phase details)
hive-create: add credential pre-check after tool discovery (Step 1A)
hive-create: add rollback/undo guidance between Steps 5 and 6
hive-patterns: add interactive AskUserQuestion pattern wizard (was passive reference doc)
hive-test: add test preview/cost-estimation step before execution (Phase 2)
hive-debugger: add Stage 6b interactive fix applier with Edit tool + user approval
settings.local.json.example: add 11 missing MCP tool permissions for debugger and test skills

Description

Brief description of the changes in this PR.

Type of Change

Bug fix (non-breaking change that fixes an issue)
New feature (non-breaking change that adds functionality)
Breaking change (fix or feature that would cause existing functionality to not work as expected)
Documentation update
Refactoring (no functional changes)

Related Issues

Fixes #(issue number)

Changes Made

Change 1
Change 2
Change 3

Testing

Describe the tests you ran to verify your changes:

Unit tests pass (cd core && pytest tests/)
Lint passes (cd core && ruff check .)
Manual testing performed

Checklist

My code follows the project's style guidelines
I have performed a self-review of my code
I have commented my code, particularly in hard-to-understand areas
I have made corresponding changes to the documentation
My changes generate no new warnings
I have added tests that prove my fix is effective or that my feature works
New and existing unit tests pass locally with my changes

Screenshots (if applicable)

Add screenshots to demonstrate UI changes.

…ebugger fix applier, test preview Fixes several workflow cross-compatibility gaps: - Add missing hive-debugger workflow stub (.agent/workflows/hive-debugger.md) - Add missing triage-issue workflow stub (.agent/workflows/triage-issue.md) - Add missing hive-debugger and triage-issue symlinks in .agent/skills/ and .agents/skills/ - Enrich all 6 existing workflow stubs with description and when_to_use metadata - hive.md: slim from ~500 to 151 lines as pure routing document (remove duplicate phase details) - hive-create: add credential pre-check after tool discovery (Step 1A) - hive-create: add rollback/undo guidance between Steps 5 and 6 - hive-patterns: add interactive AskUserQuestion pattern wizard (was passive reference doc) - hive-test: add test preview/cost-estimation step before execution (Phase 2) - hive-debugger: add Stage 6b interactive fix applier with Edit tool + user approval - settings.local.json.example: add 11 missing MCP tool permissions for debugger and test skills

Copilot

Pull request overview

This PR updates the Hive skill/workflow documentation and wiring to close cross-environment gaps, adding missing workflow stubs and enhancing several skills with interactive “wizard”-style guidance and richer metadata.

Changes:

Add missing workflow stubs (notably hive-debugger and triage-issue) and enrich workflow frontmatter (description, when_to_use) across Hive workflows.
Update multiple Hive skills with new interactive steps (patterns selection wizard, debugger fix applier stage, test execution preview, create rollback guidance).
Expand .claude/settings.local.json.example with additional MCP tool allow-list entries intended to support debugger/test workflows.

Reviewed changes

Copilot reviewed 18 out of 18 changed files in this pull request and generated 10 comments.

Show a summary per file

File	Description
.claude/skills/hive/SKILL.md	Slim down hive meta-skill into a routing/phase reference document.
.claude/skills/hive-test/SKILL.md	Add a “test preview” interaction before executing tests.
.claude/skills/hive-patterns/SKILL.md	Add an interactive AskUserQuestion-based patterns selection wizard.
.claude/skills/hive-debugger/SKILL.md	Add “Stage 6b” interactive fix application flow (Read/Edit + approval).
.claude/skills/hive-create/SKILL.md	Add credential pre-check guidance and rollback/undo guidance during graph design.
.claude/settings.local.json.example	Add additional MCP tool permissions intended for test/debug flows.
.agents/skills/triage-issue	Add skill link entry for `triage-issue`.
.agents/skills/hive-debugger	Add skill link entry for `hive-debugger`.
.agent/workflows/triage-issue.md	Add workflow stub with metadata for `triage-issue`.
.agent/workflows/hive.md	Enrich `hive` workflow metadata and add quick reference routing.
.agent/workflows/hive-test.md	Enrich `hive-test` workflow metadata.
.agent/workflows/hive-patterns.md	Enrich `hive-patterns` workflow metadata.
.agent/workflows/hive-debugger.md	Add workflow stub with metadata for `hive-debugger`.
.agent/workflows/hive-credentials.md	Enrich `hive-credentials` workflow metadata.
.agent/workflows/hive-create.md	Enrich `hive-create` workflow metadata.
.agent/workflows/hive-concepts.md	Enrich `hive-concepts` workflow metadata.
.agent/skills/triage-issue	Add skill link entry for `triage-issue`.
.agent/skills/hive-debugger	Add skill link entry for `hive-debugger`.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-19T11:03:00Z

.claude/skills/hive-test/SKILL.md

+**Before running tests, preview what will execute:**
+
+```python
+# Preview: list tests and estimate cost/time without running them
+list_tests(
+    goal_id="your-goal-id",
+    agent_path="exports/{agent_name}"
+)
+```


The comment implies list_tests can "estimate cost/time". The MCP tool only returns counts/grouping and a run command; any time/cost numbers here would be heuristic derived from the counts. Consider rewording to make it clear that the preview comes from list_tests output and any cost/time is an approximation you compute/present (not returned by the tool).

Copilot · 2026-02-19T11:03:00Z

.claude/skills/hive-test/SKILL.md

+> | Success tests | 8 | ~24s | ~$0.08 |
+> | **Total** | **12** | **~36s** | **~$0.12** |
+>
+> Each test makes real LLM calls (~3s + tokens per test).


This line states that each test makes real LLM calls. The existing guidance in this skill supports mock_mode / MOCK_MODE=1, and in that mode tests should not incur real LLM/API cost. Suggest clarifying that the cost/time estimate applies to non-mock runs (and may vary per test/tool).

Suggested change

> Each test makes real LLM calls (~3s + tokens per test).

> In non-mock runs, each test makes real LLM calls (~3s + tokens per test); with `mock_mode` / `MOCK_MODE=1`, tests run without real LLM/API cost and timing/cost estimates may differ by test/tool.

Copilot · 2026-02-19T11:03:01Z

.claude/skills/hive-create/SKILL.md

+6. **Pre-check credentials** for the discovered tools (silent — only warn if issues found):
+
+```
+check_missing_credentials(agent_path="exports/AGENT_NAME")


check_missing_credentials is an MCP tool from the agent-builder server, so this call should use the same mcp__agent-builder__... prefix used elsewhere in this step. Also, the tool loads and validates an exported agent via AgentRunner.load(agent_path), which will fail at this point because only an empty directory has been created (no agent.json yet). To make this pre-check effective, move it to after export_graph (once the agent exists), or change the approach to validate credentials based on the discovered tool list instead of an agent export path.

Suggested change

check_missing_credentials(agent_path="exports/AGENT_NAME")

mcp__agent-builder__check_missing_credentials(agent_path="exports/AGENT_NAME")

Copilot · 2026-02-19T11:03:01Z

.claude/settings.local.json.example

+      "mcp__agent-builder__query_runtime_logs",
+      "mcp__agent-builder__query_runtime_log_details",
+      "mcp__agent-builder__query_runtime_log_raw",


These permissions list query_runtime_logs, query_runtime_log_details, and query_runtime_log_raw under the agent-builder server, but those tools are registered by the separate tools MCP server (see aden_tools.tools.runtime_logs_tool). With the current entries, Claude/agents won't be able to call these tools even if tools is enabled. Update the allow-list entries to the correct server prefix (e.g., mcp__tools__query_runtime_logs, etc.).

Suggested change

"mcp__agent-builder__query_runtime_logs",

"mcp__agent-builder__query_runtime_log_details",

"mcp__agent-builder__query_runtime_log_raw",

"mcp__tools__query_runtime_logs",

"mcp__tools__query_runtime_log_details",

"mcp__tools__query_runtime_log_raw",

Copilot · 2026-02-19T11:03:01Z

.agents/skills/hive-debugger

@@ -0,0 +1 @@
+../../.claude/skills/hive-debugger


This file appears to be checked in as a regular text file containing a relative path, not as a symlink to the skill directory. Other entries under .agents/skills/ resolve as directories (symlinks) with a SKILL.md; this one does not, so the skill loader likely won't find the debugger skill. Commit this as an actual symlink (git mode 120000) pointing to ../../.claude/skills/hive-debugger (matching the existing skill link convention).

Suggested change

../../.claude/skills/hive-debugger

// (intentionally left empty; this file should be replaced by a git symlink

// pointing to ../../.claude/skills/hive-debugger)

Copilot · 2026-02-19T11:03:01Z

.agents/skills/triage-issue

@@ -0,0 +1 @@
+../../.claude/skills/triage-issue


This file appears to be checked in as a regular text file containing a relative path, not as a symlink to the skill directory. Other entries under .agents/skills/ resolve as directories (symlinks) with a SKILL.md; this one does not, so the triage-issue skill likely won't be discoverable. Commit this as an actual symlink (git mode 120000) pointing to ../../.claude/skills/triage-issue.

Suggested change

../../.claude/skills/triage-issue

Copilot · 2026-02-19T11:03:02Z

.agent/skills/hive-debugger

@@ -0,0 +1 @@
+../../.claude/skills/hive-debugger


This file appears to be checked in as a regular text file containing a relative path, not as a symlink to the skill directory. In this repo, .agent/skills/<skill> entries resolve as directories with a SKILL.md; this one does not, so the workflow stub may not be able to load the hive-debugger skill. Commit this as an actual symlink (git mode 120000) pointing to ../../.claude/skills/hive-debugger.

Copilot · 2026-02-19T11:03:02Z

.agent/skills/triage-issue

@@ -0,0 +1 @@
+../../.claude/skills/triage-issue


This file appears to be checked in as a regular text file containing a relative path, not as a symlink to the skill directory. In this repo, .agent/skills/<skill> entries resolve as directories with a SKILL.md; this one does not, so the triage-issue workflow likely won't be able to load the skill. Commit this as an actual symlink (git mode 120000) pointing to ../../.claude/skills/triage-issue.

Copilot · 2026-02-19T11:03:02Z

.claude/skills/hive-create/SKILL.md

+**To remove a node:** Call `mcp__agent-builder__remove_node(node_id="node-to-remove")` and remove its corresponding edges.
+
+**To modify a node:** Re-call `mcp__agent-builder__add_node(...)` with the same `node_id` — it overwrites the existing definition.
+
+**To remove an edge:** Call `mcp__agent-builder__remove_edge(edge_id="edge-to-remove")`.


The rollback instructions reference mcp__agent-builder__remove_node / remove_edge, but the agent-builder MCP server exposes these as delete_node and delete_edge (and delete_node already removes connected edges). Using the current names will fail at runtime; please update the tool names and guidance accordingly.

Suggested change

**To remove a node:** Call `mcp__agent-builder__remove_node(node_id="node-to-remove")` and remove its corresponding edges.

**To modify a node:** Re-call `mcp__agent-builder__add_node(...)` with the same `node_id` — it overwrites the existing definition.

**To remove an edge:** Call `mcp__agent-builder__remove_edge(edge_id="edge-to-remove")`.

**To remove a node:** Call `mcp__agent-builder__delete_node(node_id="node-to-remove")` (this will also remove any connected edges).

**To modify a node:** Re-call `mcp__agent-builder__add_node(...)` with the same `node_id` — it overwrites the existing definition.

**To remove an edge:** Call `mcp__agent-builder__delete_edge(edge_id="edge-to-remove")`.

Copilot · 2026-02-19T11:03:02Z

.claude/settings.local.json.example

+      "mcp__agent-builder__remove_node",
+      "mcp__agent-builder__remove_edge"


mcp__agent-builder__remove_node / mcp__agent-builder__remove_edge are not tools exposed by the agent-builder MCP server (the available tools are delete_node / delete_edge). As written, these permission entries won't match any callable tool name; update to the correct tool names.

Suggested change

"mcp__agent-builder__remove_node",

"mcp__agent-builder__remove_edge"

"mcp__agent-builder__delete_node",

"mcp__agent-builder__delete_edge"

Copilot AI review requested due to automatic review settings February 19, 2026 10:53

Copilot started reviewing on behalf of mangod12 February 19, 2026 10:53 View session

Copilot AI reviewed Feb 19, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Comments

docs(skills): add missing skill stubs, interactive patterns wizard, d…#5111

docs(skills): add missing skill stubs, interactive patterns wizard, d…#5111
mangod12 wants to merge 1 commit intoaden-hive:mainfrom
mangod12:feat/improve-workflow-skill-coverage

mangod12 commented Feb 19, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Feb 19, 2026

Uh oh!

Copilot AI Feb 19, 2026

Uh oh!

Copilot AI Feb 19, 2026

Uh oh!

Copilot AI Feb 19, 2026

Uh oh!

Copilot AI Feb 19, 2026

Uh oh!

Copilot AI Feb 19, 2026

Uh oh!

Copilot AI Feb 19, 2026

Uh oh!

Copilot AI Feb 19, 2026

Uh oh!

Copilot AI Feb 19, 2026

Uh oh!

Copilot AI Feb 19, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

	> Each test makes real LLM calls (~3s + tokens per test).
	> In non-mock runs, each test makes real LLM calls (~3s + tokens per test); with `mock_mode` / `MOCK_MODE=1`, tests run without real LLM/API cost and timing/cost estimates may differ by test/tool.

	check_missing_credentials(agent_path="exports/AGENT_NAME")
	mcp__agent-builder__check_missing_credentials(agent_path="exports/AGENT_NAME")

		@@ -0,0 +1 @@
		../../.claude/skills/hive-debugger No newline at end of file

	../../.claude/skills/hive-debugger
	// (intentionally left empty; this file should be replaced by a git symlink
	// pointing to ../../.claude/skills/hive-debugger)

		@@ -0,0 +1 @@
		../../.claude/skills/triage-issue No newline at end of file

		"mcp__agent-builder__remove_node",
		"mcp__agent-builder__remove_edge"

Comments

Conversation

mangod12 commented Feb 19, 2026

Description

Type of Change

Related Issues

Changes Made

Testing

Checklist

Screenshots (if applicable)

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 19, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant