scan-risks
/scan-risks
[Orbit 1.5.1 Risks] Fast risk scan: data, contracts, dependencies, rollback, observability, people. Use when: «какие риски», «что может пойти не так», «оцени риски перед стартом»; risk assessment.
- Author
- tojim00d ↗
- Repository
- Orbit-skills ↗
- Licence
- MIT
- Last repository change
- 09.11.2026
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 09.11.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
What it does and when to use it
[Orbit 1.5.1 Risks] Fast risk scan: data, contracts, dependencies, rollback, observability, people. Use when: «какие риски», «что может пойти не так», «оцени риски перед стартом»; risk assessment.
Original author description
[Orbit 1.5.1 Risks] Fast risk scan: data, contracts, dependencies, rollback, observability, people. Use when: «какие риски», «что может пойти не так», «оцени риски перед стартом»; risk assessment.
How it works
- When to use
- Inputs
- Procedure
- Output
- Check
- Pitfalls
- Portability
- What's improved vs the source
- Based on
SKILL.md preview
--- name: scan-risks description: '[Orbit 1.5.1 Risks] Fast risk scan: data, contracts, dependencies, rollback, observability, people. Use when: «какие риски», «что может пойти не так», «оцени риски перед стартом»; risk assessment.' --- # scan-risks — fast risk scan of a task > **Orbit · ① Understand · step 1.5.1 Risks** > **Input:** the task and recon — from `survey-codebase`, `model-domain`, `clarify-task` > **Output:** `.orbit/task/<id>/risks.md` → next `model-threats` > **Works with:** `clarify-task`, `write-spec`, `plan-steps` > **Shared rules:** reply in the user's language; generated artifacts default to English; artifacts go to `.orbit/` (contract: `ARTIFACTS.md`). Fifteen minutes before work that save days. Not only security (that is `model-threats`) — all risks: technical, data, deadlines, people, operations. ## When to use - A task is about to start and the blast radius is not written down. - A schema change, API change or deploy is planned and rollback is unproven. - Dependencies, quotas or access rights are unclear and could block the work. - Another team's release window, freeze or on-call rotation can affect delivery. - Before `write-spec` or `plan-steps`, so risks become tasks instead of surprises. Do **not** turn it into a security review (that is `model-threats`) and do not produce a risk register for a typo fix. ## Inputs - `.orbit/task/<id>/brief.md` — goal, done-criteria, scope; - `.orbit/knowledge/overview.md`, codebase state, the project map — what the change touches; - `git log --oneline -20` and the current branch state — recent churn and in-flight work; - deploy and infra files: Dockerfile, Helm values, CI workflows, migrations; - `.orbit/task/<id>/risks.md` if the scan continues — append, do not replace. If the recon is missing, read the five files the change touches and say the scan is partial. If the owner of a risk is unknown, that is itself a risk to record. ## Procedure 1. **Restate the change in one line** and the system area it touches. The scan is about that change, not about the codebase. 2. **Walk the checklist below and mark only real risks** — one line per area that is genuinely affected. Empty areas are skipped, not filled with "no risk". 3. **Check data risk in the migrations.** Schema change? Backfill? Reversible? Volume? Personal data? ```bash find . -path '*/migrations/*' -newer .orbit/task/<id>/brief.md | head grep -rnE 'ALTER TABLE|DROP COLUMN|ADD COLUMN .*NOT NULL' db/ migrations/ 2>/dev/null | head ``` 4. **Check contracts.** API, event or file-format change — who consumes it? Is there versioning? A contract change without a consumer list is the highest-impact risk. ```bash grep -rn '<endpoint-or-event>' --include='*.ts' --include='*.py' --include='*.go' . | grep -v node_modules | head -20 ``` 5. **Check dependencies.** New library or service: licence, maintenance, availability in the internal registry. A new dependency is a supply-chain and an approval risk. 6. **Check infrastructure.** Resources, quotas, network policies, secrets in the vault, access rights. In k8s, render the chart before assuming the values. ```bash helm template <release> ./chart -f values-prod.yaml > /tmp/rendered.yaml && grep -nE 'resources:|limits:|replicas:' /tmp/rendered.yaml | head ``` 7. **Check rollback and observability.** One-command rollback? What happens to data on rollback? Will a failure be seen — metrics, alerts, logs? ```bash git log --oneline -10 -- <deploy-dir> grep -rn 'health\|readiness\|metrics' <deploy-dir> | head ``` 8. **Check unknowns and people.** What did we assume instead of checking in code? Who is needed from other teams, change windows and freezes? 9. **Rate and rank.** Likelihood (low/med/high) × impact (low/med/high); sort by score. Top 3 risks get an action and an owner. 10. **Write the artifact and wire it in.** Top 3 risks go into `proposal.md` (`run-spec-loop`) and into the plan (`plan-steps`) as separate tasks: "check X before starting", "prepare rollback". A risk that is an unchecked assumption → ask now (`clarify-task`, `think-first`). | Area | Questions | |---|---| | Data | Schema change? Migration? Reversible? Volume? Personal data? | | Contracts | API/event/format change? Who consumes it? Versioning? | | Dependencies | New library/service? Licence? Maintained? Available in the internal registry? | | Infrastructure | Resources, quotas, network policies, secrets in the vault, access rights? | | Rollback | One-command rollback? What happens to data on rollback? | | Observability | Will we see a failure? Metrics, alerts, logs? | | Unknowns | What did we assume instead of checking in code? | | People and time | Who's needed from other teams? Change windows, freezes? | ## Output `.orbit/task/<id>/risks.md` (shared with `model-threats`): ```markdown # Risks — <task> — <date> ## Scan | Risk | Likelihood | Impact | Score | Action | Owner | |---|---|---|---|---|---| | backfill on 40M rows locks orders | M | H | 6 | batch backfill + dry run | @data | | report consumer breaks on new field | M | M | 4 | add field, keep old one | @export | | no rollback for migration 007 | L | H | 6 | reversible down + tested restore | @platform | ## Assumptions to check - [ ] retention policy covers the new column — owner @legal ## Top 3 → plan tasks / proposal 1. … ``` ## Check ```bash test -f .orbit/task/<id>/risks.md && echo "artifact ok" grep -cE '^\| .* \| (L|M|H) \| (L|M|H) \|' .orbit/task/<id>/risks.md # rated risks grep -n 'Top 3' .orbit/task/<id>/risks.md grep -cE '^- \[ \]' .orbit/task/<id>/risks.md # unchecked assumptions git status --short | grep -E 'risks|plan' || true ``` Pass rule: only real risks are listed (no filler rows); each has likelihood, impact and an owner; the top 3 appear as tasks in `plan.md` or items in `proposal.md`; unchecked assumptions have a checkbox and an owner; the scan names which inputs were missing. Link the file from `brief.md` so later steps read it. ## Pitfalls - **A risk list nobody acts on.** A row without an owner or a plan task is decoration; the top 3 must exist in `plan.md`. - **Scanning security twice.** That is `model-threats`; this scan covers data, contracts, dependencies, rollback, observability, people — security is just one row if relevant. - **Assuming the rollback works.** "We can revert the deploy" ignores data migrations. Test the down path or record the gap as a top risk. - **Missing the consumer list.** A contract change discovered after release is the most expensive version of this scan; grep for consumers before rating it. - **Filling empty areas.** "No risks in people" adds noise and hides the real ones. Skip unaffected areas. - **One-time scan.** Re-run at plan review and before deploy; the risk set changes when the design does. ## Portability Markdown, `git` and `grep` — any agent and model can do it, and it fits in a 15-minute timebox. Without subagents, walk the areas in order; the artifact is the shared state. If `helm` or the deploy tooling is unavailable, inspect the manifests directly and note the fallback. No Claude Code features are required. ## What's improved vs the source - Not only security: data, contracts, rollback, observability, people and change windows in one 15-minute checklist. - Top 3 risks become plan tasks and proposal items, not a forgotten note. - Every risk has an owner, and unchecked assumptions are explicit checkboxes. - The scan starts from the change and its blast radius (`trace-impact`), not from a generic questionnaire. ## Based on - Orbit's own skill.
Versions
The database has no other skills named scan-risks.
Database snapshot: 13.11.2026 · package version 3493194d3aaf (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 100model-threats
[Orbit 1.5.2 Risks] STRIDE threat model over data flows before code; controls → spec requirements. Use when: «модель угроз», «STRIDE», «безопасность до кода», «что может пойти не так с безопасностью»; threat model. - 90risk-assessment
Identify, assess, and mitigate operational risks. Trigger with "what are the risks", "risk assessment", "risk register", "what could go wrong", or when the user is evaluating risks associated with a project, vendor, process, or decision. - 100risk-scan
Identifies, evaluates, and recommends mitigation strategies for project risks across any phase of delivery. Use whenever a PM needs continuous risk discovery or decision support - including when someone says "analyse the risks on this project", "do a full risk analysis", "we… - 90risk-radar
Surfaces project risks — timeline slippage signals, resource concerns, scope creep warnings, dependency risks, and stakeholder concerns — buried in project email threads. Use when a project manager wants an early warning scan of risks before they become problems. Triggers on… - 90hr-risk-management
Help HR teams identify, assess, and mitigate people-related risks across compliance, workforce continuity, conduct, and organizational health. Use when asked to identify HR risks, build an HR risk register, manage compliance risk, develop HR risk mitigation plans, handle… - 100risk-register-builder
Create a structured risk register for any project or initiative with identified risks, probability/impact scoring, risk owners, mitigation plans, and contingency actions. Use when the user says 'build a risk register', 'identify project risks', 'risk assessment', 'what could go…