| name | hone:automation-opportunities |
| description | Finds manual processes that could be automated: manual deploy steps in READMEs, setup instructions that could be scripts, repetitive git workflows, manual testing procedures, and undocumented tribal knowledge. Also checks for artifacts not in source control. Use monthly to surface automation ROI. Do NOT use for code quality or style audits. |
| compatibility | Requires a local repository checkout with documentation, scripts, CI config, and source files accessible on disk. Works with any technology stack. |
| metadata | {"short-description":"Finds manual processes ripe for automation"} |
Automation Opportunities
What This Skill Does
Audits the repository for manual processes, undocumented workflows,
and artifacts that should be in source control. Produces a prioritized
list of automation opportunities with estimated effort and impact.
Detection categories:
- Manual deploy steps: multi-step deployment instructions in
READMEs, wikis, or runbooks that could be a single script or CI
pipeline.
- Setup scripts missing: setup instructions that list manual
commands instead of providing a bootstrap script.
- Repetitive git workflows: branching, tagging, or release
processes described as manual steps.
- Manual testing procedures: test plans or checklists that could
be automated test suites or CI checks.
- Tribal knowledge in comments: comments explaining "how to do X"
that should be executable scripts or documented runbooks.
- Missing from source control: referenced config files, env
templates, tool configs, database migrations, or infrastructure
definitions that are mentioned but not checked in.
When To Use
- Monthly scheduled audit to surface high-ROI automation.
- When onboarding is slow and you suspect undocumented manual steps.
- After an incident caused by a manual process failure.
- When planning a DevEx improvement sprint.
Do Not Use
- For code quality, style, or naming audits (use other hone skills).
- For security scanning.
- For performance analysis.
- As a replacement for a full DevOps maturity assessment.
Inputs To Confirm
- Scope -- which directories and doc files to scan
(default: entire repo including docs, READMEs, CI configs, and
Makefiles).
- Focus areas -- whether to prioritize deploy automation, setup
automation, or all categories (default: all).
- Team size context -- rough team size to help estimate
automation ROI (default: not specified, omit ROI estimates).
Instructions
- Identify the repository root and enumerate all files, including
documentation (README, CONTRIBUTING, docs/), CI/CD configs
(.github/workflows, .gitlab-ci.yml, Jenkinsfile, etc.), scripts
directories, Makefiles, Dockerfiles, and package manager configs.
- Manual deploy steps: scan READMEs, docs, and runbooks for
deployment instructions. Flag when:
- There are 3+ sequential manual shell commands for deploying.
- Instructions include "SSH into" or "run on the server" steps.
- The deploy process references manual environment variable
setting, config file copying, or service restarts.
- There is no CI/CD pipeline config, or the pipeline does not
cover the documented deploy steps.
- Setup scripts missing: scan for setup or getting-started
instructions. Flag when:
- README lists 5+ manual commands to get the project running.
- There is no
setup.sh, bootstrap, make setup, init
script, or equivalent.
- Instructions reference manual tool installation without a
version manager or Dockerfile.
- Repetitive git workflows: scan for branching, release, or
versioning documentation. Flag when:
- Release steps involve manual version bumping, changelog editing,
and tag creation.
- There is no release automation (semantic-release, standard-version,
etc.).
- Branch naming conventions are documented but not enforced by
hooks or CI.
- Manual testing procedures: scan for test plans, QA checklists,
or manual verification steps. Flag when:
- There are documented manual test cases that could be automated.
- README or CONTRIBUTING mentions manual browser testing without
an E2E test suite.
- There are "before deploying, verify that..." checklists.
- Tribal knowledge in comments: scan source files for comments
that describe operational procedures:
- "To regenerate this file, run..."
- "This must be updated whenever..."
- "Ask [person] about how to..."
- "The trick is to..."
Flag these as candidates for executable scripts or runbooks.
- Missing from source control: check for references to files or
configs that should exist but do not:
.env.example or .env.template referenced but missing.
- Config files mentioned in docs but not in the repo.
- Infrastructure-as-code references without corresponding files.
- Database migration or seed files mentioned but absent.
- Tool config files (
.editorconfig, linter configs) mentioned
in CONTRIBUTING but not present.
- For each finding, record:
- Category (one of the six above).
- Location: file path and line number or section reference.
- Description of the manual process or missing artifact.
- Impact estimate:
high (repeated frequently, error-prone, or
blocks onboarding), medium (done occasionally, moderate risk),
low (rare but worth automating eventually).
- Effort estimate:
small (< 1 day), medium (1-3 days),
large (> 3 days).
- Suggested automation approach (e.g., "add a Makefile target",
"create a GitHub Action", "write a bootstrap script").
- Produce the output report.
Output Requirements
Produce a Markdown report with:
- Summary: total findings, breakdown by category and impact,
and a one-paragraph assessment of the repo's automation maturity.
- Findings by category: one section per category with a table of
findings (location, description, impact, effort, suggested
approach).
- ROI ranking: findings sorted by impact/effort ratio -- high
impact and small effort first.
- Quick wins: the top 3-5 findings that deliver the most value
for the least effort.
- Missing artifacts checklist: a simple checklist of files that
should be added to source control.
Quality Bar
- Every finding must cite a specific file, line, or documentation
section -- not a vague "you should automate deploys."
- Impact and effort estimates must be grounded in observable evidence
(e.g., "the deploy doc has 12 manual steps" justifies high impact).
- The report must distinguish between missing automation and
automation that exists but is incomplete.
- Suggestions must be concrete (name a specific tool, script type, or
CI feature) rather than generic advice.
- The missing artifacts checklist must reference the document or
comment that mentions the missing file.