/contract-reviewLegal
Reads a contract portfolio against your own clause standard and escalates only what breaches it.
142 agreements audited in one run, 3 uncapped liability clauses sent to the partner.SkillsSkills library
A skill is a saved workflow with a slash command in front of it. The library is what your organisation has already agreed is the right way to do something, made available to everyone who needs it, running the same way whoever starts it. Twenty-four of the most-used are below; the catalogue runs past 270.
The library
Every skill below carries the result line from a real run. The numbers are what that run produced, not a ceiling and not a promise.
Reads a contract portfolio against your own clause standard and escalates only what breaches it.
142 agreements audited in one run, 3 uncapped liability clauses sent to the partner.Sorts a data room by relevance and privilege before anyone bills an hour reading it.
9,400 documents triaged overnight, 61 flagged as privileged and withheld.Puts the same clause from every agreement with a counterparty side by side and marks the drift.
Seven master agreements reconciled to one position paper in under an hour.Turns a matter file into a chronology with every date carried back to the document that fixes it.
A four-year matter reduced to a 3-page chronology, every entry cited.Matches positions across the ledger, the trading system and the collateral book.
142 variances raised, each traced back to the source row that caused it.Rebuilds counterparty exposure from the systems that hold it rather than from last month’s extract.
Exposure restated across 9 entities without a single manual export.Runs the month-end checklist against the actual state of the ledger and reports what is genuinely outstanding.
Close cycle shortened by two days; 14 open items surfaced on day one.Explains a budget variance by walking back to the transactions that produced it.
Every line over threshold explained with its own transaction list attached.Re-screens a book of clients against sanctions, PEP and adverse media, and writes the file note.
2,800 clients refreshed, 37 escalated with the evidence already assembled.Reads a policy against the regulation it implements and lists what the policy does not cover.
Nine gaps identified against the updated rulebook, each quoted from both sides.Pulls the evidence for a control test from the systems of record and assembles the working paper.
A quarterly control test evidenced without a single screenshot request.Reconstructs who accessed what and when, from the logs, into a timeline an investigator can use.
A 72-hour window reconstructed across five systems into one ordered account.Builds the pre-meeting brief from your own history with the account and what is publicly known now.
Every first meeting opens with the last four years of the relationship in hand.Reads usage, support history and sentiment together and ranks the book by what is genuinely at risk.
Eleven renewals reprioritised six weeks before anyone would have noticed.Drafts the response from the answers your team has already given, and flags what is genuinely new.
A 240-question RFP drafted to 81% from prior approved answers.Reads closed opportunities and call transcripts together and reports the pattern, not the anecdote.
Two consistent loss reasons isolated across 300 closed opportunities.Assembles the post-incident account from monitoring, tickets and chat, in order, with sources.
Post-incident reviews written the same day rather than the following week.Puts a supplier’s contractual position, spend and delivery record on one page.
Three suppliers found billing above the contracted rate card.Reads demand and the current run rate and produces a capacity plan with the method shown.
Plan rebuilt monthly instead of quarterly, with the model visible each time.Traces what a proposed change touches across the estate before anyone approves it.
Two undocumented downstream dependencies found before the change window.Turns a job description and the performance record for the role into a defensible scorecard.
Interview panels working from one standard rather than five.Reads the comp band, the market position and the internal spread together, per team.
Pay review evidenced against internal equity as well as market data.Maps a market from filings, coverage and your own record, and says where the sources disagree.
A market map that names its own uncertainty instead of hiding it.Reads the literature on a question and reconciles the findings, citation by citation.
Sixty papers reduced to one position, every claim carried back to its source.24 shown · 270+ in the catalogue. The library above is the most-used slice. Anything in the wider catalogue can be enabled for your tenant, and your own skills sit alongside them. Nothing matching? Skills are written, not found, so if the work exists the skill can too — write your own.
Your own skills
The generic ones are a starting point. The valuable ones encode how your firm does something, which nobody outside it can write for you.
A skill is authored by the person who knows the work, reviewed like anything else that touches production, and then available to the whole team behind one command. When the standard changes, it changes in one place.
In plain language: the inputs, the standard it is judged against, and what a finished result looks like.
Which connectors it may read, and which writes it is allowed to make.
It is versioned, diffed and approved. Nobody edits the live one by accident.
It appears in the library for the roles you scope it to, with the same audit trail as everything else.
Skills run on demand or on a schedule. If the work should happen without anyone starting it, see scheduled tasks. If you want the prompt rather than the whole workflow, the prompt library is the lighter end of the same idea.
Bring the workflow your team runs most often and we will show you it as a skill.