Find tasks in a work plan that each wait on the other

A draft cycle report listing each prerequisite cycle and invalid reference, with alternative sequencing questions.

Choose one of two approaches. The steps for that option appear below.

Choose an option above. On a phone, swipe the tabs left or right to see all options →

ChatGPT: finds the cycles with code, drafts the report

ChatGPT builds the edge list with readable Python, then drafts the cycle report. You check every edge and the report.

  1. You

    You: Get source pack

  2. ChatGPT

    ChatGPT: Build edge list

  3. You

    You: Check every edge

  4. ChatGPT

    ChatGPT: Draft cycle report

  5. You

    You: Completion check

  1. Get the source pack ready

    Use: You

    Have ready: the task IDs, prerequisite edges and milestone gates. Paste the pack into one ChatGPT chat and keep the original files open.

    Check

    Check nothing beyond this review is included and every original stays open for comparison.

  2. Build the edge list with cycles

    Use: ChatGPT

    Ask ChatGPT to build the edge list: every prerequisite edge with its source reference, invalid references flagged, circular dependencies listed, with readable Python code.

    Details

    What to paste into the tool

    Task: Draft edge list with cycles and invalid references with source IDs, dates, uncertainty markers and named gaps. Using only the material I supplied, build a validated edge list: one row per prerequisite edge with its source reference, flagging edges that reference task IDs or gates not in the supplied material, and a list of every circular dependency. Do the cycle detection and reference checks with Python code I can read, and show me the code with the output. Copy source identifiers, dates and quotes exactly as written. Where the material is silent or ambiguous, write 'not stated' and note the gap - do not invent a value. Inputs: [paste the source pack, or upload it as files].

  3. Check every edge row

    Use: You

    Check every row of the edge list against the original sources; ask for corrections in the same chat.

    Check

    Compare identifiers, definitions, dates, quotes and ambiguous claims against the original inputs.

  4. Draft the cycle report

    Use: ChatGPT

    In the same chat, ask ChatGPT to draft the cycle report from the checked edge list, with alternative sequencing questions and no schedule changes.

    Details

    What to paste into the tool

    Task: Draft cycle report with alternative sequencing questions; Unresolved questions with source references. From the checked table and the supplied sources, draft a cycle report that describes each cycle with the task IDs and edges involved and their source references, plus alternative sequencing questions for the planner, without changing any schedule. Keep every source reference, keep each unresolved item open as a question with its source reference, and add no new facts. If any check is needed, do it with Python code I can read and show the code. Inputs: [paste the checked edge list with reviewer corrections].

  5. Run the completion check

    Use: You

    Check each cycle against the true prerequisites; no schedule is changed. The workflow ends at this checked draft.

    Check

    Check each cycle against true prerequisites; nothing is sent, published, assigned or booked. Decide any next action separately.

Before you paste into any AI tool, remove any confidential information. Put it back at the end.

Tools in this workflow

Final check

Every option checks the first table against the original sources before anything moves on.

Every path ends at a reviewed draft under the source check; nothing is sent, published, assigned, reserved, booked, bought or changed.

Every path keeps unresolved items open with source references instead of inventing values.

Check the result against your original inputs. Approve it before sharing or acting.

What still needs checking
  • Plan access: which plan tiers (if any) the reviewer holds for the named tools was not checked, and no option assumes a paid plan. Where a vendor page gates a feature to a paid tier (for example the Gemini overview describes 'recently uploaded files' as '(Gemini Advanced only)'), the evidence plan_qualification says to confirm on the account first (the Gemini upload step text carries no separate plan caveat).
  • Every verdict marked 'not yet checked' is unproven until the human verification steps in the path have run.

Share this page