decision-handler

Name: decision-handler
Author: KnoxOps

38stars4forks

How to install decision-handler?

knoxops/open-devops-skills/decision-handler

$npx -y skills add KnoxOps/open-devops-skills --skill decision-handler

Installs into the current project.

Summary

Final phase of the zombie resource cleanup workflow. Presents all Phase 2 findings (deep scan + observation) to the user for final human decision, executes confirmed deletions, and delivers a cost savings report.

Files · 1

View on GitHub

SKILL.md

1	## Input Parameters
2
3	\| Name \| Type \| Required \| Description \|
4	\|------\|------\|----------\|-------------\|
5	\| run_dir \| string \| Yes \| Workspace root directory for file passing between agents \|
6	\| task_id \| string \| Yes \| Task ID for progress tracking \|
7	\| ssh_key_path \| string \| No \| SSH key path for cloud provider access \|
8
9	## Execution Flow
10
11	### Task Context
12
13	Before starting execution, initialize `task_context.json`:
14
15	```json
16	{
17	"task_id": "<task_id from input>",
18	"current_step": 0,
19	"current_step_id": null,
20	"status": "running",
21	"steps": {
22	"load_and_merge": "pending",
23	"generate_decision_report": "pending",
24	"write_phase5_decision": "pending",
25	"execute_decisions": "pending",
26	"deliver_final_report": "pending"
27	},
28	"updated_at": "<ISO timestamp>"
29	}
30	```
31
32	Update this file after each step completes. On error, set step status to `"failed"` and overall `status` to `"failed"`.
33
34	### Step 1: load_and_merge
35
36	Type: agent
37	Description: Load all Phase 2 outputs and merge into a unified per-resource view.
38	Sorted by cost descending for prioritization.
39
40
41	## Input Files
42	- `analysis/deep_scan_{resource_id}.json`
43	- `observe/observation_{resource_id}.json`
44	- `observe/observation_batch_summary.json`
45
46	## Execution
47	Launch an independent agent with the following prompt file:
48
49	Dispatch instruction:
50
51	Load and merge all Phase 2 outputs into a single unified view.
52
53	## Inputs to read
54
55	1. Deep scan profiles - glob `{run_dir}/analysis/deep_scan_*.json`
56	Each file contains detailed technical profile for one candidate resource:
57	- resource_id, resource_type, entity_type, cloud_provider
58	- spec (CPU/memory/disk), actual_utilization
59	- running_processes, open_ports, installed_packages
60	- owner_detail, team_lead_id
61	- business_context (application, system, scheduled_tasks)
62	- cost (monthly, yearly - either from cloud billing or estimated)
63	- tags, environment
64
65	2. Observation reports - glob `{run_dir}/observe/observation_*.json`
66	Each file contains isolation observation results:
67	- resource_id
68	- isolation_method, isolation_start/end dates
69	- observation_result: passed / failed / uncertain
70	- alert_count, complaint_count
71	- anomaly_events (if any), rollback_triggered
72	- observation_period_days
73
74	3. Observation batch summary - `{run_dir}/observe/observation_batch_summary.json`
75	High-level summary of all observations (results histogram, rollback count).
76
77	## Merge logic
78
79	For each unique resource_id found across deep scan files:
80	- If both deep_scan + observation exist: merge full profile
81	- If only deep_scan exists (observation not yet done): mark observation_result = "pending"
82	- If only observation exists (deep scan missing): include with available data, mark spec = "incomplete"
83
84	## Recommendation logic
85
86	For each resource, generate an AI recommendation:
87	- DELETE: observation_result = "passed" (no alerts, no complaints, no rollback)
88	- KEEP: observation_result = "failed" OR resource has active business use
89	- EXTEND_OBSERVATION: observation_result = "uncertain" (some alerts but inconclusive)
90
91	This is a preliminary recommendation - the human makes the final call.
92
93	Sort resources by cost.monthly descending (most expensive first).
94
95	## Self-validation
96
97	Before writing the output, verify:
98	- total_candidates matches resources array length
99	- All resources have required fields: resource_id, resource_type, cloud_provider, cost, observation, recommendation, recommendation_reason
100	- total_monthly_cost equals sum of all resources monthly cost
101	- observation_summary counts equal total_candidates
102	- Resources sorted by cost.monthly descending
103
104	If any validation fails, fix the issue and re-validate.
105
106
107	Agent workflow:
108
109	1. Read input data from:
110	- Schema: `schemas/deep-scan.schema.json`
111	- Schema: `schemas/observation-input.schema.json`
112	- Schema: `schemas/observation-batch-summary-input.schema.json`
113
114	2. Execute the agent with the prompt
115
116	3. Write results to:
117	- File: `decision/merged_view.json`
118
119	## Output
120	- Schema: schemas/merged-view.schema.json
121	- File: decision/merged_view.json
122
123	### Progress Tracking
124
125	After completing this step, update `task_context.json`:
126	- Set `current_step_id` to `"load_and_merge"`
127	- Set `steps.load_and_merge` to `"completed"`
128	### Step 2: generate_decision_report
129
130	Type: agent
131	Description: Present merged findings to the user and collect per-resource decisions.
132	Use the Write tool to produce a Markdown report in the chat.
133
134
135	## Input Files
136	- `decision/merged_view.json` (from Step load_and_merge, schema: schemas/merged-view.schema.json)
137
138	## Execution
139	Launch an independent agent with the following prompt file:
140
141	Prompt file: `$P

People who liked this also viewed

6 picks

Type

decision-handler

38stars4forks

View on GitHub

How to install decision-handler?

knoxops/open-devops-skills/decision-handler

$npx -y skills add KnoxOps/open-devops-skills --skill decision-handler

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	## Input Parameters
2
3	\| Name \| Type \| Required \| Description \|
4	\|------\|------\|----------\|-------------\|
5	\| run_dir \| string \| Yes \| Workspace root directory for file passing between agents \|
6	\| task_id \| string \| Yes \| Task ID for progress tracking \|
7	\| ssh_key_path \| string \| No \| SSH key path for cloud provider access \|
8
9	## Execution Flow
10
11	### Task Context
12
13	Before starting execution, initialize `task_context.json`:
14
15	```json
16	{
17	"task_id": "<task_id from input>",
18	"current_step": 0,
19	"current_step_id": null,
20	"status": "running",
21	"steps": {
22	"load_and_merge": "pending",
23	"generate_decision_report": "pending",
24	"write_phase5_decision": "pending",
25	"execute_decisions": "pending",
26	"deliver_final_report": "pending"
27	},
28	"updated_at": "<ISO timestamp>"
29	}
30	```
31
32	Update this file after each step completes. On error, set step status to `"failed"` and overall `status` to `"failed"`.
33
34	### Step 1: load_and_merge
35
36	Type: agent
37	Description: Load all Phase 2 outputs and merge into a unified per-resource view.
38	Sorted by cost descending for prioritization.
39
40
41	## Input Files
42	- `analysis/deep_scan_{resource_id}.json`
43	- `observe/observation_{resource_id}.json`
44	- `observe/observation_batch_summary.json`
45
46	## Execution
47	Launch an independent agent with the following prompt file:
48
49	Dispatch instruction:
50
51	Load and merge all Phase 2 outputs into a single unified view.
52
53	## Inputs to read
54
55	1. Deep scan profiles - glob `{run_dir}/analysis/deep_scan_*.json`
56	Each file contains detailed technical profile for one candidate resource:
57	- resource_id, resource_type, entity_type, cloud_provider
58	- spec (CPU/memory/disk), actual_utilization
59	- running_processes, open_ports, installed_packages
60	- owner_detail, team_lead_id
61	- business_context (application, system, scheduled_tasks)
62	- cost (monthly, yearly - either from cloud billing or estimated)
63	- tags, environment
64
65	2. Observation reports - glob `{run_dir}/observe/observation_*.json`
66	Each file contains isolation observation results:
67	- resource_id
68	- isolation_method, isolation_start/end dates
69	- observation_result: passed / failed / uncertain
70	- alert_count, complaint_count
71	- anomaly_events (if any), rollback_triggered
72	- observation_period_days
73
74	3. Observation batch summary - `{run_dir}/observe/observation_batch_summary.json`
75	High-level summary of all observations (results histogram, rollback count).
76
77	## Merge logic
78
79	For each unique resource_id found across deep scan files:
80	- If both deep_scan + observation exist: merge full profile
81	- If only deep_scan exists (observation not yet done): mark observation_result = "pending"
82	- If only observation exists (deep scan missing): include with available data, mark spec = "incomplete"
83
84	## Recommendation logic
85
86	For each resource, generate an AI recommendation:
87	- DELETE: observation_result = "passed" (no alerts, no complaints, no rollback)
88	- KEEP: observation_result = "failed" OR resource has active business use
89	- EXTEND_OBSERVATION: observation_result = "uncertain" (some alerts but inconclusive)
90
91	This is a preliminary recommendation - the human makes the final call.
92
93	Sort resources by cost.monthly descending (most expensive first).
94
95	## Self-validation
96
97	Before writing the output, verify:
98	- total_candidates matches resources array length
99	- All resources have required fields: resource_id, resource_type, cloud_provider, cost, observation, recommendation, recommendation_reason
100	- total_monthly_cost equals sum of all resources monthly cost
101	- observation_summary counts equal total_candidates
102	- Resources sorted by cost.monthly descending
103
104	If any validation fails, fix the issue and re-validate.
105
106
107	Agent workflow:
108
109	1. Read input data from:
110	- Schema: `schemas/deep-scan.schema.json`
111	- Schema: `schemas/observation-input.schema.json`
112	- Schema: `schemas/observation-batch-summary-input.schema.json`
113
114	2. Execute the agent with the prompt
115
116	3. Write results to:
117	- File: `decision/merged_view.json`
118
119	## Output
120	- Schema: schemas/merged-view.schema.json
121	- File: decision/merged_view.json
122
123	### Progress Tracking
124
125	After completing this step, update `task_context.json`:
126	- Set `current_step_id` to `"load_and_merge"`
127	- Set `steps.load_and_merge` to `"completed"`
128	### Step 2: generate_decision_report
129
130	Type: agent
131	Description: Present merged findings to the user and collect per-resource decisions.
132	Use the Write tool to produce a Markdown report in the chat.
133
134
135	## Input Files
136	- `decision/merged_view.json` (from Step load_and_merge, schema: schemas/merged-view.schema.json)
137
138	## Execution
139	Launch an independent agent with the following prompt file:
140
141	Prompt file: `$P

People who liked this also viewed

6 picks

Type